Majid Jaberi-Douraki

Background/Objectives: Fluoroquinolones are classified by the World Health Organization as highest-priority critically important antimicrobials, and their use in food-producing animals is a One Health concern. We aimed to map the published literature on fluoroquinolones in food-producing animals and to extract the dosing regimens and reports of resistance it contains. Methods: We screened 77,583 Scopus records (1981–2021) and applied natural language processing, combining named-entity recognition with rule-based pattern matching, to 14,417 abstracts on cattle, pigs, poultry, sheep, and goats. Extraction was validated against a blinded manual annotation of 205 records, and temporal trends and between-species differences were assessed with Mann–Kendall and Kruskal–Wallis tests. Results: Extraction achieved an overall precision of 0.80, recall of 0.70, and F1-score of 0.75. Annual publications increased steadily (Mann–Kendall τ = 0.98, p < 0.001), led by authors affiliated with institutions in the United States, China, and India. Across 531 doses from 370 records, the median reported dose was 10 mg/kg in poultry and 5 mg/kg in cattle, pigs, sheep, and goats (p < 0.001); 67% of doses came from pharmacokinetic or residue studies. The proportion of records reporting fluoroquinolone resistance rose from 13% in the 1990s to about 20% after 2000 and has since remained stable, most often for ciprofloxacin with Escherichia coli, Campylobacter, and Salmonella. Conclusions: Natural language processing can map dosing and resistance information in the veterinary literature at scale. These outputs describe research activity and reported regimens, not fluoroquinolone consumption or resistance prevalence, and are best interpreted alongside national sales and surveillance data.

Clearance (CL) is a primary pharmacokinetic (PK) parameter crucial to determine how quickly a drug is eliminated from the body, which guides the appropriate dosing interval to maintain a consistent concentration in blood. Given the importance of CL, this study aimed to use machine learning (ML) techniques to predict CL values by identifying patterns and relationships within an extracted dataset of PK variables from published articles. Variables evaluated in the extracted dataset included drug, dose, animal species, and route of administration. Nine distinct ML models were then applied to analyze the CL data, incorporating both imbalanced and balanced data generated through resampling methods. Since the CL data used in this study is a collection of all CL values (true CL and CL/F) extracted from scientific articles, the collected CL variable for both IV and non-IV administration routes are referred to as hybrid ML CL. To analyze the effect of ML models in predicting the CL values, we used the hybrid ML CL dataset for six different subsets of data including one solely from the intravenous route of administration. Linear regression, multi-layer perceptron, and random forest models consistently had the highest efficiency in predicting CL values, with an R ² score > 0.87. However, R ² increased to > 0.95 when analyzing only ungulates or small ruminants, and > 0.92 for the companion animal group. This study has the potential to help researchers employ computational, mathematical, and ML models to predict and estimate CL values and changes in CL values based on variables. This study focuses on evaluating the feasibility of predicting drug CL in situations where direct CL data are not available. Rather than addressing drug development processes, the research examines whether study design variables can serve as input parameters for a proposed cross-species extrapolation tool aimed specifically at predicting existing drug CL values.

The extraction and standardization of pharmacokinetic (PK) information from scientific literature remain significant challenges in computational pharmacology, which limits the reliability of data-driven models in drug development. Large language models (LLMs) have achieved remarkable progress in text understanding and reasoning, yet their adaptation to structured biomedical data, such as PK tables, remains constrained by heterogeneity, noise, and domain shift. To address these limitations, we propose HySim-LLM, a unified mathematical and computational framework that integrates embedding-weighted fine-tuning and manifold-aware denoising to enhance the robustness and interpretability of LLMs. We establish two theoretical results: (1) a similarity-weighted generalization bound that quantifies adaptation performance under embedding divergence, and (2) a manifold-based denoising guarantee that bounds loss contributions from noisy or off-manifold samples. These theorems provide a principled foundation for fine-tuning LLMs in structured biomedical settings. The framework offers a mathematically grounded pathway toward reliable and interpretable LLM adaptation for biomedical and data-intensive scientific domains.

The safe use of pharmaceuticals in food-producing animals is vital to protect animal welfare and human food safety. Adverse events (AEs) may signal unexpected pharmacokinetic or toxicokinetic effects, increasing the risk of violative residues in the food chain. This study introduces a predictive framework for classifying outcomes (Death vs. Recovery) using ~1.28 million reports (1987-2025 Q1) from the U.S. FDA's OpenFDA Center for Veterinary Medicine. A preprocessing pipeline merged relational tables and standardized AEs through VeDDRA ontologies. Data were normalized, missing values imputed, and high-cardinality features reduced; physicochemical drug properties were integrated to capture chemical-residue links. We evaluated supervised models, including Random Forest, CatBoost, XGBoost, ExcelFormer, and large language models (Gemma 3-27B, Phi 3-12B). Class imbalance was addressed, such as undersampling and oversampling, with a focus on prioritizing recall for fatal outcomes. Ensemble methods(Voting, Stacking) and CatBoost performed best, achieving precision, recall, and F1-scores of 0.95. Incorporating Average Uncertainty Margin (AUM)-based pseudo-labeling of uncertain cases improved minority-class detection, particularly in ExcelFormer and XGBoost. Interpretability via SHAP identified biologically plausible predictors, including lung, heart, and bronchial disorders, animal demographics, and drug physicochemical properties. These features were strongly linked to fatal outcomes. Overall, the framework shows that combining rigorous data engineering, advanced machine learning, and explainable AI enables accurate, interpretable predictions of veterinary safety outcomes. The approach supports FARAD's mission by enabling early detection of high-risk drug-event profiles, strengthening residue risk assessment, and informing regulatory and clinical decision-making.

Bispecific antibodies (BsAbs) have shown promise in the management of relapsed/refractory multiple myeloma (MM). Despite its efficacy, this class of drugs is associated with significant toxicities. In this study, we conducted a pooled analysis of the available clinical trials on BsAbs for the treatment of MM, including full publications and abstracts until April 2025. BsAbs were classified into two groups: B-cell maturation antigen (BCMA), and GPRC5D/FcRH5 BsAbs. Welch’s t-test was performed to compare the safety profiles of each agent. For clustering, we used principal component analysis (PCA). Our study analyzed 22 trials involving 2374 patients with MM from early 2023 to April 2025. Among these, 1276 patients received BCMA BsAbs, 841 treated with GPRC5D/FcRH5 BsAbs, 157 received teclistamab + talquetamab, and 65 patients received a talquetamab + daratumumab, and 35 patients received talquetamab + pomalidomide. The median follow-up for all groups was 11.83 months. Among all-grade hematologic adverse events (AEs), neutropenia occurred in 40.4%, anemia in 39.2%, thrombocytopenia in 21.4%, lymphopenia in 19.2%, infections in 45.8%, and cytokine release syndrome (CRS) in 65%. For grade 3/4 AEs, infections occurred in 20.3%, CRS in 1.5%, neutropenia in 35.2%, anemia in 24.5%%, thrombocytopenia in 13.5%, and lymphopenia in 17.7%. CRS and the need for tocilizumab were significantly less frequent with BCMA BsAbs vs GPRC5D/FcRH5 BsAbs, (P < 0.002). Skillings Mack (Generalized Friedman’s) findings emphasized substantial distinctions between BCMA and GPRC5D/FcRH5×CD3 in both overall and severe grade 3/4 AEs (p ≤ 0.0002). PCA revealed agents with all grades and grade 3/4 showed similar clustering patterns except for three agents. Overall, our findings demonstrated the excellent efficacy on the use of BsAbs in MM; however, these agents have been linked to a unique AE profile. GPRC5D/FcRH5 are associated with less grade 3/4 hematologic toxicity whereas BCMA BsAbs were associated with lower grade 3/4 CRS rates, compared to GPRC5D/FcRH5. These insights are crucial for guiding treatment decisions and developing strategies to improve patient outcomes.

Multiple myeloma (MM) is a complex hematological malignancy of clonal plasma cells driven by alterations to the chromosomal material leading to uncontrolled proliferation in the bone marrow. Ethnic and racial disparities persist in the prevalence, diagnosis, management, and outcomes of MM. These disparities are multifaceted and intersect with various factors, including demographics, geography, socioeconomic status, genetics, and access to healthcare. This study utilized the openFDA human drug adverse events (AEs) to analyze global data pertaining to MM patients and patterns of treatment-related AEs. We identified ten most frequently used drugs and drug regimens in six distinct regions, including North America (NA), Europe (EU), Asia (AS), Africa (AF), Oceania (OC), and Latin America & the Caribbean (LA). AE patterns were evaluated using the reporting odds ratio combined with a 95% confidence interval. AE reports were more prevalent in men than in women across all regions. Cardiotoxicities were more likely observed in AS and EU, while secondary neoplasms were more frequently reported in the EU. Nephropathies were prominent in OC, AF (in males), and AS (in females), while vascular toxicity, including embolism and thrombosis, was more common in NA (in males). A notable improvement in survival, particularly in AS, EU, and NA, with a significant decline in death rates was observed. Hospitalization rates displayed less variation in AS and EU but exhibited more pronounced fluctuations in AF, LA, and OC. In conclusion, this comprehensive analysis offers valuable insights into the demographic, geographic, and AE patterns of MM patients across the globe.

Ossabaw pigs (n = 11; 5—gilts, 6—barrows; age 15.6 ± 0.62 SD months) were exposed to a three-choice preference maze to evaluate preference for fermented sorghum teas (FSTs). After conditioning, pigs were exposed, in four sessions, to choices of white FST, sumac FST, and roasted sumac-FST. Then, pigs were exposed, in three sessions, to choices of deionized H₂O (−control; avoidance), isocaloric control (+control; deionized H₂O and sucrose), and blended FST (3Tea) (equal portions: white, sumac, and roasted sumac). When tea type was evaluated, no clear preference behaviors for tea type were observed (p > 0.10). When the 3Tea and controls were evaluated, pigs consumed minimal control (p < 0.01;18.0 ± 2.21% SEM), and they consumed great but similar volumes of +control and 3Tea (96.6 and 99.0 ± 2.21% SEM, respectively). Likewise, head-in-bowl duration was the least for −control, but 3Tea was the greatest (p < 0.01; 5.6 and 31.9 ± 1.87% SEM, respectively). Head-in-bowl duration for +control was less than 3Tea (p < 0.01; 27.6 vs. 31.9 ± 1.87% SEM). Exploration duration was the greatest in the area with the −control (p < 0.01; 7.1 ± 1.45% SEM), but 3Tea and +control exploration were not different from each other (1.4 and 3.0 ± 1.45% SEM, respectively). Regardless of tea type, adult pigs show preference for FST, even over +control. Adult pigs likely prefer the complexity of flavors, rather than the sweetness alone.

Three-choice preferences for fermented sorghum teas (FST) were examined among 11 adult Ossabaw pigs. The objectives for experiments 1 and 2 (respectively) were to determine preference for: 1) one of three types of FST (white vs. sumac vs. roasted sumac), and 2) isocaloric control (+control; made isocaloric with sucrose), blended FST (3Tea; equal parts of white, sumac, and roasted-sumac), or a negative control (-control; distilled water; avoidance). Pigs were first conditioned to navigate a three-choice preference maze. For experiment 1, no clear preference behaviors for tea-type were observed. Pigs consumed nearly all the teas, but pigs consumed the least amount of tea in the first session. Pigs spent the most time with their heads in the bowl when 3Tea was offered, followed by +control, and the least time was -control. Regardless of tea-type, adult pigs show a strong preference for FST, even over +control, which was sweetened water. Adult pigs likely prefer the complexity of flavors, rather than the sweetness alone. Future projects that explore the health benefits of this sorghum formula will not be impeded by avoidance among adult pigs.