Background: The global burden of dengue infection has rising, yet limited data exists on its impact in the Caribbean. We describe the incidence and associates of acute kidney injury in adults and children with dengue at a teaching hospital in Jamaica. Methods: A single-centre retrospective cohort study of admissions with laboratory confirmed dengue infection at University Hospital of the West Indies, Mona Jamaica between January 2023 to November 2024. AKI was defined using Kidney Disease Improving Global Outcomes definitions. Patients were included if aged >1year and had at least 2 creatinine values. Clinical, demographic and laboratory data were abstracted by chart review. Summary statistics were used to describe continuous and categorical data, and logistic regression to determine AKI associations. Stratified analysis was performed by age-group (adults-aged [≥] 16, and paediatric-aged <16 years). Results: Analyses included 167 persons, 62% (103) were male, mean age was 26.1+/-19.5 years. AKI occurred in 25.8%, 65.1% were KDIGO stage 1. AKI incidence was 30.2% and 18.0% among adults and children respectively. There were 3 in-hospital deaths. People with AKI were older 32+/-21.4 vs 24 +/-18.4 (p=0.021), and had longer duration of stay [6 vs 4 days (p <0.001)]. Male sex [OR 2.09 (95% CI:0.96-4.59), p=0.064], age per year [OR 1.02 (95% CI:1.01-1.04), p=0.015] symptom duration [OR1.11 (CI 0.99-1.24), p = 0.058], admission bilirubin [OR 1.02 (CI: 1.00-1.04), p = 0.022], NLR [OR 1.09 (CI 1.00-1.18), p = 0.037] were associated with AKI. In adults admission potassium was inversely associated with AKI [OR 0.46 (95% CI 0.21-1.01), p 0.056], while in children admission potassium [OR 3.00 (95% CI 0.88-10.6), p 0.088] was associated with AKI. Conclusion: AKI in dengue hospitalizations is higher than most reports at 25.8%. Targeted public health policy on vector control and early symptom recognition may be needed to improve outcomes.
Depression subtype classification from social media posts: few-shot prompting vs. fine-tuning of large language models
BackgroundSocial media provides timely proxy signals of mental health, but reliable tweet-level classification of depression subtypes remains challenging due to short, noisy text, overlapping symptomatology,




