Machine learning (ML) algorithms are increasingly applied in medical research and in healthcare, gradually improving clinical practice. Among various applications of these novel methods, their usage in the combat against antimicrobial resistance (AMR) is one of the most crucial areas of interest, as increasing resistance to antibiotics and management of difficult-to-treat multidrug-resistant infections are significant challenges for most countries worldwide, with life-threatening consequences. As antibiotic efficacy and treatment options decrease, the need for implementation of multimodal antibiotic stewardship programs is of utmost importance in order to restrict antibiotic misuse and prevent further aggravation of the AMR problem. Both supervised and unsupervised machine learning tools have been successfully used to predict early antibiotic resistance, and thus support clinicians in selecting appropriate therapy. In this paper, we reviewed the existing literature on machine learning and artificial intelligence (AI) in general in conjunction with antimicrobial resistance prediction. This is a narrative review, where we discuss the applications of ML methods in the field of AMR and their value as a complementary tool in the antibiotic stewardship practice, mainly from the clinician’s point of view.

In patients with acute respiratory failure, gas exchange is impaired due to the accumulation of fluid in the lung airspaces. This life-threatening syndrome is treated with mechanical ventilation, which is adjusted to maintain gas exchange, but can be associated with the accumulation of carbon dioxide in the lung. Carbon dioxide (CO₂) is a by-product of cellular energy utilization and its elimination is affected via alveolar epithelial cells. Signaling pathways sensitive to changes in CO₂ levels were described in plants and neuronal mammalian cells. However, it has not been fully elucidated whether non-neuronal cells sense and respond to CO₂. The Na,K-ATPase consumes ∼40% of the cellular metabolism to maintain cell homeostasis. Our study examines the effects of increased pCO₂ on the epithelial Na,K-ATPase a major contributor to alveolar fluid reabsorption which is a marker of alveolar epithelial function. We found that short-term increases in pCO₂ impaired alveolar fluid reabsorption in rats. Also, we provide evidence that non-excitable, alveolar epithelial cells sense and respond to high levels of CO₂, independently of extracellular and intracellular pH, by inhibiting Na,K-ATPase function, via activation of PKCζ which phosphorylates the Na,K-ATPase, causing it to endocytose from the plasma membrane into intracellular pools. Our data suggest that alveolar epithelial cells, through which CO₂ is eliminated in mammals, are highly sensitive to hypercapnia. Elevated CO₂ levels impair alveolar epithelial function, independently of pH, which is relevant in patients with lung diseases and altered alveolar gas exchange.

Linezolid is a new antimicrobial agent effective against drug-resistant gram-positive pathogens commonly responsible for central nervous system (CNS) infections in neurosurgical patients hospitalized in intensive care units. In order to study the penetration of this antimicrobial into the cerebrospinal fluid (CSF) of such patients, the disposition of linezolid in serum and CSF was studied in 14 neurosurgical patients given linezolid at 600 mg twice daily (1-h intravenous infusion) for the treatment of CNS infections caused by gram-positive pathogens or for prophylactic chemotherapy. Serum and CSF linezolid steady-state concentrations were analyzed by high-pressure liquid chromatography, and the concentration-time profiles obtained were analyzed to estimate pharmacokinetic parameters. The mean ± standard deviation (SD) linezolid maximum and minimum measured concentrations were 18.6 ± 9.6 μg/ml and 5.6 ± 5.0 μg/ml, respectively, in serum and 10.8 ± 5.7 μg/ml and 6.1 ± 4.2 μg/ml, respectively, in CSF. The mean ± SD areas under the concentration-time curves (AUCs) were 128.7 ± 83.9 μg · h/ml for serum and 101.6 ± 59.6 μg · h/ml for CSF, with a mean penetration ratio for the AUC for CSF to the AUC for serum of 0.66. The mean elimination half-life of linezolid in CSF was longer than that in serum (19.1 ± 19.0 h and 6.5 ± 3.6 h, respectively). The serum and CSF linezolid concentrations exceeded the pharmacodynamic breakpoint of 4 μg/ml for susceptible target pathogens for the entire dosing interval in the majority of patients. These findings suggest that linezolid may achieve adequate concentrations in the CSF of patients requiring antibiotics for the management or prophylaxis of CNS infections caused by gram-positive pathogens.

Ventilator-associated pneumonia (VAP) remains a common risk in mechanically ventilated patients. Different care bundles have been proposed to succeed VAP reduction. We aimed to identify the combined interventions that have been used to by ICUs worldwide from the implementation of “Institute for Healthcare Improvement Ventilator Bundle”, i.e., from December 2004. A search was performed on the PubMed, Scopus and Science Direct databases. Finally, 38 studies met our inclusion criteria. The most common interventions monitored in the care bundles were sedation and weaning protocols, semi-recumbent positioning, oral and hand hygiene, peptic ulcer disease and deep venus thrombosis prophylaxis, subglottic suctioning, and cuff pressure control. Head-of-bed elevation was implemented by almost all studies, followed by oral hygiene, which was the second extensively used intervention. Four studies indicated a low VAP reduction, while 22 studies found an over 36% VAP decline, and in ten of them, the decrease was over 65%. Four of these studies indicated zero or nearly zero after intervention VAP rates. The studies with the highest VAP reduction adopted the “IHI Ventilator Bundle” combined with adequate endotracheal tube cuff pressure and subglottic suctioning. Multifaced techniques can lead to VAP reduction at a great extent. Multidisciplinary measures combined with long-lasting education programs and measurement of bundle’s compliance should be the gold standard combination.