Background: Clostridioides difficile infection (CDI) is a major cause of healthcare-associated infections globally. Understanding variations in CDI incidence and outcomes across settings, populations, and regions is important for guiding prevention strategies. Aim: The aim of this study was to determine the global epidemiology of CDI to better understand disease burden across settings and geographic regions. Methods: Relevant publications were identified through searches of major databases, including PubMed, Scopus, and Web of Science, published from 1 January 2016 through 24 July 2024. Random effects models were used to pool estimates, and 95% confidence intervals (CIs) were calculated. Results: A total of 59 studies, representing 24 countries across North America, Europe, the Asia–Pacific region, Latin America, and the Middle East, met the inclusion criteria. The incidence of CDI was highest in hospital-onset healthcare facility settings, with 5.31 cases/1000 admissions (95% CI 3.76–7.12) and 5.00 cases/10,000 patient-days (95% CI 3.96–6.15). Long-term care facilities reported 44.24 cases/10,000 patient-days (95% CI 39.57–49.17). Pediatric populations faced a greater risk, with 4.52 cases/1000 admissions (95% CI 0.55–12.17), than adults did at 2.13 (95% CI 1.69–2.61). Recurrence rates were highest for community-acquired CDI at 16.22%. The death rates for the CDI cases tracked for 30 days and of unspecified duration were 8.32% and 16.05%, respectively. Conclusions: This comprehensive review identified healthcare facilities, long-term care, pediatric populations, and North America as disproportionately burdened. This finding provides guidance on priority areas and populations for targeted prevention through antimicrobial stewardship, infection control, and surveillance.

Multidrug-resistant Klebsiella pneumoniae (MDR-Kp) poses a significant public health challenge due to limited treatment options and growing antimicrobial resistance (AMR). The rise in MDR-Kp infections underscores the critical need for new strategies against antibiotic resistance, especially in Africa, where AMR is a pressing concern. This systematic review evaluated the antimicrobial activity of herbal preparations against MDR-Kp by assessing parameters such as minimum bactericidal concentration (MBC), zone of inhibition, and minimum inhibitory concentration (MIC). A literature search was done following the PRISMA guidelines, and relevant articles from PubMed, Scopus, and ScienceDirect were included. The search identified 2094 articles, of which 848 were selected for a full assessment and 31 were included. Data extraction included information on plant scientific names, botanical families, plant parts, country of origin, extract types, isolated compounds, and antimicrobial parameters. Of the 124 plants studied, Withania frutescens L., Plumbago zeylanica, Plectranthus glandulosus, Eruca sativa, and Donella welwitschii, emerged as plants that could be further explored for anti-MDR-Kp therapeutics. Further research is needed to explore their mechanisms of action, active compounds, and potential synergies. Integrating traditional knowledge with modern healthcare practices, standardised research methodologies, and clinical validations can advance plant-based therapies. Additionally, conservation efforts are vital to preserve the biodiversity of these medicinal plants, ensuring their availability for future research and therapeutic use, while supporting environmental and public health.

Frontier large language models (LLMs) trained with extended reasoning, such as DeepSeek-R1 and OpenAI o1, have established state-of-the-art benchmarks in formal logic, competitive programming, and mathematics. In interactive, empirical scientific sandboxes, however, their efficacy remains counter-intuitive. In the Bioinformatics Benchmark (BixBench), extended reasoning models paradoxically degraded in multi-turn environments, leading benchmark authors to exclude them from primary agentic evaluations. In this paper, we conduct the first systematic forensic audit of the internal cognitive monologue ( ... ) of open-weight reasoning models (DeepSeek-R1-Distill-Qwen-14B) and matched instruction-tuned controls (Qwen2.5-Coder-14B) deployed inside containerized computational biology sandboxes on Nvidia H100 GPUs. Our forensic analysis uncovers two distinct systemic failure modes: (1) The PubMed Prior Trap, in which reasoning models engage in epistemic rationalization within their cognitive stream, falsely convincing themselves that empirical datasets are missing and substituting empirical computation with speculative literature recall (e.g., estimating adjusted p-values based on "typical experimental outcomes"); and (2) Tag Language Nesting Collisions, in which models conflate agent-level XML action tags with Python syntax, producing terminal interpreter syntax errors. To resolve this paradox, we introduce Dual Track Decoupled Reasoning (DTDR), an agent scaffolding architecture that evicts internal cognitive monologues from multi-turn history, injects strict empirical grounding guardrails, and dynamically intercepts runtime tool invocations. On a stratified diagnostic benchmark spanning single-cell RNA-seq, bulk differential expression (DESeq2), and genomic variant prioritization (n=22 per arm), DTDR rescues the 14B reasoning model from 0.0% to 9.09% accuracy (2/22), doubling the performance of the matched non-reasoning coder control (4.55%, 1/22) while slashing code execution errors by 81% (13.6% vs. 72.7%) and reducing average steps to solution by 66% (1.77 vs. 5.18 steps). All replication pipelines, prompts, and trajectory manifests are openly released.