Vascular access is the major risk factor for bacteremia, hospitalization, and mortality among hemodialysis (HD) patients. The type of vascular access most associated with bloodstream infection is central venous catheter (CVC). The incidence of catheter-related bacteremia ranges between 0.6 and 6.5 episodes per 1000 catheter days and increases linearly with the duration of catheter use. Given the high prevalence of CVC use and its direct association with catheter-related bacteremia, which adversely impacts morbidity and mortality rates and costs among HD patients, several prevention measures aimed at reducing the rates of CVC-related infections have been proposed and implemented. As a result, a large number of clinical trials, systematic reviews, and meta-analyses have been conducted in order to assess the effectiveness, clinical applicability, and long-term adverse effects of such measures. In the following article, prophylactic measures against CVC-related infections in HD patients and their possible advantages and limitations will be discussed, and the more recent literature on clinical experience with prophylactic antimicrobial lock therapy in HD CVCs will be reviewed.

Background: Catheter-related bloodstream infection (CR-BSI) is one of various complications related to hemodialysis (HD). As a result of the high rate of infection, the use of lock solutions for the prevention of CR-BSI has been studied. However, adverse effects of lock solution, such as increased emergence of strains resistant to antibiotics, which is an important concern, need to be investigated further. The aim of this study was to compare the efficacy of lock solution using a combination of cefazolin and gentamicin versus taurolidine and citrate in reducing CR-BSI in patients undergoing HD and to identify any adverse effects. Methods: A prospective observational study was performed at two dialysis centers. Patients using new tunneled central venous catheters (CVC) for HD were included. Patients with a tunneled CVC were assigned to receive either antibiotic lock solution (group 1: gentamicin 7 mg/ml + cefazolin 12 mg/ml + heparin 3500 IU/ml) or lock solution with TauroLock-Hep500 (group 2: taurolidine citrate 4% + heparin 500 IU/ml) during the inter-dialysis period. The patients were allocated to these groups according to the hemodialysis center they were attending. Results: A total of 145 CVCs were implanted in 127 patients and were followed for 15 months: 77 CVCs (65 patients) were placed in group 1 and 68 CVCs (62 patients) in group 2. There was no difference between the two groups with regard to CR-BSI (events per 1000 catheter-days: group 1 = 0.79, group 2 = 1.10; p = 0.18) or exit site infection rates (events per 1000 catheter-days: group 1 = 2.45, group 2 = 1.83; p = 0.37). The groups differed in ESI pathogens, with gram-positive oxacillin-resistant pathogens more frequent in group 1 (31.8% vs. 5.0%; p = 0.003). The two groups were similar in mechanical complications. In the Cox regression analysis, the internal jugular vein site was a protective factor for all catheter removal complications (hazard ratio (HR) 0.41, 95% confidence interval (CI) 0.19-0.91) and mechanical complications (HR 0.16, 95% CI 0.065-0.41); only ESI was a risk factor for all catheter removal complications (HR 1.79, 95% CI 1.04-3.07) and mechanical complications (HR 5.64, 95% CI 1.65-19.3). Conclusions: The efficacy of both lock solutions was similar in preventing infections related to tunneled CVCs for HD. However, there were more oxacillin-resistant strains in patients who received antibiotic lock solution. Further studies are required to determine the optimal drug regimen and concentrations for lock solution and the associated adverse effects.

In the applied sciences, the ultimate goal is not just to acquire knowledge but to turn knowledge into action. The next wave for data disciplines may be experimental designs and analytical methods for closing the gap between the "real-world" situations faced by decision-makers and their idealized representations in optimization problems, and the health sciences are poised to be the discipline where these developments substantially improve lives. We discuss three recent trends in research-experimental designs and analytical methods for precision medicine and pragmatic trials; technological developments in sensors, wearables, and smartphones for measuring health data; and methods addressing algorithmic bias and model interpretability-and argue that these seemingly disparate trends point to a future where data-driven decision support tools are increasingly used to promote wellbeing.

One of the key steps in building deep learning systems for drug classification and generation is the choice of featurization for the molecules. Previous featurization methods have included molecular images, binary strings, graphs, and SMILES strings. This paper proposes the creation of molecular images "captioned" with binary vectors that encode information not contained in or easily understood from a molecular image alone. Specifically, we use Morgan fingerprints, which encode higher level structural information, and MACCS keys, which encode yes/no questions about a molecule's properties and structure. We tested our method on the HIV dataset published by the Pande lab, which consists of 41,127 molecules labeled by if they inhibit the HIV virus. Our final model achieved a state-of-the-art AUC-ROC on the HIV dataset, outperforming all other methods. Moreover, the model converged significantly faster than most other methods, requiring dramatically less computational power than unaugmented images.