Search

Viet Tran Quoc,1 Dung Nguyen Thi Ngoc,2,3 Trung Nguyen Hoang,4 Hoa Vu Thi,4 Minh Tong Duc,4 Thanh Do Pham Nguyet,2 Thanh Nguyen Van,5 Diep Ho Ngoc,2 Giang Vu Son,6 Thanh Bui Duc7 1Intensive Care Unit, Military Hospital 175, Ho Chi Minh City, Vietnam; 2Department of Military Science and Training, Military Hospital 175, Ho Chi Minh City, Vietnam; 3Hanoi University of Public Health, Hanoi, Vietnam; 4Department of Military Hygiene, Vietnam Military Medical University, Hanoi, Vietnam; 5Department of General Planning, Military Hospital 175, Ho Chi Minh City, Vietnam; 6Department of Personnel, Military Hospital 175, Ho Chi Minh City, Vietnam; 7Institute of Trauma and Orthopedics, Military hospital 175, Ho Chi Minh City, VietnamCorrespondence: Thanh Bui Duc, Military Hospital 175, 786 Nguyen Kiem, 3 Ward, Go Vap District, Ho Chi Minh City, Vietnam, Tel +84 975 430 258, Email buiducthanh175@yahoo.comIntroduction: Artificial Intelligence (AI) and machine learning (ML) are used extensively in HICs to detect and control antibiotic resistance (AMR) in laboratories and clinical institutions. ML is designed to predict outcome variables using an algorithm to enable “machines” to learn the “rules” from the data. ML is increasingly being applied in intensive care units to identify AMR and to assist empiric antibiotic therapy. This study aimed to evaluate the performance of ML models for predicting AMR bacteria and resistance to antibiotics in two Vietnamese hospitals.Patients and Methods: A cross-sectional study combined with retrospective was conducted from 1st January 2020 to 30th June 2022. Five models were developed to predict antibiotic resistance of bacterial infections of ICU patients. Two datasets were prepared to predict AMR bacteria and antibiotics with ML models. The performance of the prediction models was evaluated by various indicators (sensitivity, specificity, precision, accuracy, F1-score, PRC, AuROC, and NormMCC) to determine the optimal time point for data selection. Python version 3.8 was used for statistical analyses.Results: The accuracy, F1-score, AuROC, and normMMC of LightGBM, XGBoost, and Random Forest models were higher than those of other models in both datasets. In both datasets 1 and 2, accuracy, F1-score, AuROC and normMCC of the XGBoost model were the highest among five models (from 0.890 to 1.000). Only Random Forest models had specificity scores higher than 0.850. High scores of sensitivity, accuracy, precision, F1-score, and normMCC indicated that the models were making accurate predictions for datasets 1 and 2.Conclusion: XGBoost, LightGBM, and Random Forest were the best-performed machine learning models to predict antibiotic resistance of bacterial infections of ICUs patients using the patients’ EMRs.Keywords: antibiotic resistance, machine learning, XGBoost, LightGBM, random forest

To reveal post-traumatic stress disorder (PTSD) genetic risk influences on tissue-specific gene expression, we use brain and non-brain transcriptomic imputation. We impute genetically regulated gene expression (GReX) in 29,539 PTSD cases and 166,145 controls from 70 ancestry-specific cohorts and identify 18 significant GReX-PTSD associations corresponding to specific tissue-gene pairs. The results suggest substantial genetic heterogeneity based on ancestry, cohort type (military versus civilian), and sex. Two study-wide significant PTSD associations are identified in European and military European cohorts; ZNF140 is predicted to be upregulated in whole blood, and SNRNP35 is predicted to be downregulated in dorsolateral prefrontal cortex, respectively. In peripheral leukocytes from 175 marines, the observed PTSD differential gene expression correlates with the predicted differences for these individuals, and deployment stress produces glucocorticoid-regulated expression changes that include downregulation of both ZNF140 and SNRNP35. SNRNP35 knockdown in cells validates its functional role in U12-intron splicing. Finally, exogenous glucocorticoids in mice downregulate prefrontal Snrnp35 expression.

This paper presents an experimental study on how the crack-parallel stress affects the fracture properties of fiber-reinforced concrete (FRC) using the gap test—a new simple fracture test invented and used for concrete at Northwestern University in 2020. First, it was conducted for plain concrete and was successfully applied to cross-ply carbon-fiber composite and to aluminum. An advantage of this test is that it is unambiguous because the test setup changes from one statically determinate configuration to another. The gap test, combined with the standard notched three-point-bend test, is now applied to geometrically scaled FRC specimens to determine how the fracture energy, Gf , and the effective size, cf , of the fracture process zone (FPZ), are changed by the crack-parallel stress, σxx. For σxx equal to about 2/3 of the standard uniaxial compression strength, the increase in Gf is 64% and 78% for the two FRCs, respectively, which is large but not as large as the 126% increase observed in tests of plain concrete. This indicates that the fiber reinforcement mitigates the effect of σxx , while introducing some degree of ductility into the fracture process. The compressive σxx also increases the effective size of the FPZ by about 81% and 64% while such increase is 134% in plain concrete. Because crack-parallel stresses are ubiquitous in practice, the implications for design are significant. [ABSTRACT FROM AUTHOR]

In the existing database on shear load capacity, tests of very large beams are scarce. Valuable additions to the database have recently been made in 2021 at the University of California, Berkeley (UCB), and in 2015 at the University of Toronto. These two tests were the largest ever among the standard three-point-bend type tests conducted so far. They verified the effects of beam size and of steel stirrups on the ultimate load, Vu , provided that the same concrete and steel are used. The present analysis, which deals in detail only with the UCB test, shows that the subsequent public blind competitions to predict the Vu measured in both tests were meritless and potentially misleading. The reason is that, similar to design codes, the only information provided to the competitors (besides the E modulus) was the required concrete compression strength, fc′ , whereas the mean compressive and tensile strengths, fracture energy, initial creep data, and so on, were not provided. The fault of a competition of this kind is evidenced by (1) finite-element fracture simulations, (2) analysis of the huge statistical scatter of a database of 784 tests and a previous database in which fc′ was also the only concrete property used, like in the design code, and (3) estimation of the statistical error due to anchoring code provisions to the classical shear strength approximation 2fc′ (psi), which was set at about 65% below the mean of the data cloud in the database. The winning prediction of the UCB competition had an error of only 2.7% of the measured failure load, even though the probability of success is here shown to have been between 0.14% and 8.46%, with 0.90% being the best estimate. Hence, competitions of this type are, in essence, a lottery. Furthermore, the fact that the winning predictions in both competitions happened to be obtained by cross-section strain analysis based on beam mechanics, and no fracture mechanics, is potentially misleading. This, of course, does not detract from the value of the UCB and Toronto experiments as important and unique additions to the database and as verifications of the load capacity for the particular concrete used. [ABSTRACT FROM AUTHOR]