Background: GPT-4 is an extensive language model designed to understand and generate human-like text, with the ability to accomplish complex tasks such as data analysis and decision-making. Methods: We conducted a retrospective observational study, which involved preprocessing 340 clinical cases, the contents of which included medical history, physical findings, laboratory, and instrumental data. We provided this information to GPT-4, which was instructed to give us the top five most likely differential diagnoses. The discrepancies between artificial intelligence (AI)’s five differentials and the physician’s final diagnosis were analyzed regarding different specialties and laboratory/instrument data inclusion. Results: Before we integrated laboratory data, GPT-4 showed a 60% diagnostic match with physicians on the first differential and 86% in all five. After we integrated the instrumental and laboratory findings, these percentages grew to 72% and 92%, respectively. The integration of laboratory and instrumental data increased GPT-4’s diagnostic accuracy, evidenced by an odds ratio factor of 2.1 and McNemar’s test chi-squared value of 10.76, highlighting the substantial impact of this data on AI’s diagnostic precision. Conclusions: This study highlighted the potential of using GPT-4 in medical diagnostics, provided a basis for integrating AI tools with clinical judgment, and opened avenues for future research, particularly in developing AI models tailored to the diagnostic needs of different medical fields.

Acute respiratory distress syndrome (ARDS) is a severe, life-threatening condition characterized by acute hypoxemic respiratory failure, bilateral pulmonary infiltrates, and non-cardiogenic pulmonary edema caused by increased alveolar-capillary permeability. ARDS is highly heterogeneous, with diverse etiologies and clinical presentations that complicate diagnosis and management. Viral infections, including influenza A (H1N1 and H5N1) and coronaviruses such as SARS-CoV-2, are major contributors to ARDS and can trigger severe lung injury, hyperinflammation, and dysregulated immune responses. Ongoing viral evolution and periodic emergence of novel strains continue to pose a substantial threat to global public health. This narrative review analyzes pandemic-associated viral causes of ARDS, summarizes key mechanisms of pathogenesis, and evaluates current and emerging pharmacotherapeutic approaches. A comprehensive literature search was conducted using PubMed, supplemented by additional sources where appropriate. The review highlights that the increasing prominence of viral pneumonia as a cause of ARDS requires both established supportive care and tailored therapeutic strategies that target the underlying mechanisms of lung injury. Despite progress, virus-associated ARDS remains a major clinical challenge with high morbidity and mortality and may require management approaches distinct from those used for other ARDS etiologies.

Introduction: Differentiating candidemia from bacteremia in intensive care unit (ICU) patients is critical for effective treatment but is often delayed by reliance on conventional diagnostics like blood cultures. This study evaluates the diagnostic utility of interleukin-6 (IL-6), C-reactive protein (CRP), and soluble triggering receptor expressed on myeloid cells ⁻¹ (sTREM-1), individually and in combination, for distinguishing these bloodstream infections (BSIs). Methods: This prospective observational study included 74 ICU patients with BSIs admitted to a university-affiliated hospital between September 2023 and August 2024. Biomarker levels were measured on the 2 ^nd day of ICU admission. Diagnostic performance was assessed using the Mann–Whitney U test, logistic regression, and receiver operating characteristic curve analysis. Results: Of the 74 patients, 54 (72.9%) had bacteremia and 20 (27.1%) had candidemia. IL-6 and sTREM-1 levels were significantly higher in candidemia ( P < 0.001), whereas CRP alone did not show discriminatory power ( P = 0.981). The combined biomarker model achieved an area under the curve of 0.818, demonstrating superior diagnostic performance compared to individual markers. Conclusions: IL-6 and sTREM-1 are effective biomarkers for distinguishing candidemia from bacteremia, with CRP providing additional value in a multibiomarker model. This approach may enhance diagnostic precision and improve clinical decision-making in ICU settings.

Galectin-3 has been recognized as a biomarker of inflammation and fibrosis in COVID-19, yet its association with echocardiographic indicators of cardiac involvement remains insufficiently understood. This study aimed to assess the relationship between galectin-3 levels and echocardiographic changes of the left ventricle and left atrium in hospitalized patients with COVID-19. A retrospective analysis was conducted on 41 adults who were divided into moderately symptomatic patients (21/41, 51.2%; within the first week of symptom onset) and severely symptomatic patients (20/41, 48.8%; during the second to third week). All patients underwent standard laboratory testing, including galectin-3 measurement, and transthoracic echocardiography upon admission. Galectin-3 levels were significantly higher in the severe group compared with the moderate group (13.7 ± 7.9 ng/mL vs. 8.7 ± 6.2 ng/mL; p = 0.029). Severe illness was also associated with a larger ascending aortic diameter (p = 0.041) and greater left atrial size (p = 0.05). Although interventricular septal thickness, posterior wall thickness, and left ventricular end-diastolic diameter were similar between groups, left ventricular end-diastolic volume, end-systolic volume, and left ventricular mass were higher in severe cases, but these differences did not reach statistical significance. Left ventricular ejection fraction remained normal in moderate cases and was mildly reduced in severe cases without a statistically significant difference. Galectin-3 demonstrated significant correlations with several echocardiographic markers, most notably left atrial diameter and indices of left ventricular structure and function. These findings suggest that elevated galectin-3 levels are associated with structural myocardial changes in COVID-19. Combined assessment of galectin-3 and echocardiographic parameters may aid in identifying early myocardial involvement and improving cardiovascular risk stratification in hospitalized COVID-19 patients.

Background: Handwashing is a critical measure for reducing diarrhoeal and respiratory infections among school-aged children, yet maintaining consistent practice remains a challenge. Behavioural factors, including daily routines, environmental cues, and perceptions of hygiene spaces, shape handwashing behaviour beyond the availability of water and soap. This study examined the behavioural determinants of handwashing among rural schoolchildren in Georgia using the Capability, Opportunity, Motivation-Behaviour (COM-B) model. Methods: A cross-sectional survey was conducted in 2024 among 368 students from 10 rural public schools in the Adjara region. A structured interviewer-administered questionnaire assessed sociodemographic characteristics, knowledge, environmental conditions, motivational factors, and self-reported handwashing practices. Adequate handwashing was defined as washing hands both before eating and after toilet use. Variables associated with adequate behaviour in bivariate analyses ( p < 0.20) were considered for inclusion in a multivariable logistic regression model. Results: Overall, 85.9% of students reported adequate handwashing behaviour. Although knowledge of correct handwashing steps was high (91.3%), substantial behavioural gaps were observed. In the adjusted model, three factors independently predicted adequate handwashing: not relying solely on washing when hands were visibly dirty—indicating a stable hygiene routine (AOR = 30.3, 95% CI [13.0–71.4]); perceiving the school toilet environment as comfortable or safe (AOR = 6.6, 95% CI [1.8–23.8]); and washing hands at least five times per day (AOR = 2.22, 95% CI [1.04–4.74]). Water, adequate Sanitation, and proper Hygiene (WASH) education showed a borderline association (AOR = 2.15, p = 0.060). Knowledge-related indicators and availability of water or soap were not significant predictors in the multivariable analysis. Conclusion: Handwashing behaviour among rural schoolchildren in Georgia is driven less by knowledge or basic facility availability and more by routine formation, perceived comfort of hygiene spaces, and motivational factors. Strengthening handwashing habits, improving the qualitative aspects of school toilets, and incorporating behaviourally informed approaches into hygiene education may enhance the sustainability of WASH outcomes. Applying the COM-B model provides guidance for designing interventions that support consistent, protective hygiene practices in school settings.

Functional independence in children with cerebral palsy (CP) varies with motor involvement and development. We describe two-year changes in Functional Inde