DKADE: a novel framework based on deep learning and knowledge graph for identifying adverse drug events and related medications.

Adverse drug events (ADEs) are common in clinical practice and can cause significant harm to patients and increase resource use. Natural language processing (NLP) has been applied to automate ADE detection, but NLP systems become less adaptable when drug entities are missing or multiple medications are specified in clinical narratives. Additionally, no Chinese-language NLP system has been developed for ADE detection due to the complexity of Chinese semantics, despite ˃10 million cases of drug-related adverse events occurring annually in China. To address these challenges, we propose DKADE, a deep learning and knowledge graph-based framework for identifying ADEs. DKADE infers missing drug entities and evaluates their correlations with ADEs by combining medication orders and existing drug knowledge. Moreover, DKADE can automatically screen for new adverse drug reactions. Experimental results show that DKADE achieves an overall F1-score value of 91.13%. Furthermore, the adaptability of DKADE is validated using real-world external clinical data. In summary, DKADE is a powerful tool for studying drug safety and automating adverse event monitoring.

Comparative effectiveness of Lopinavir/Ritonavir-based regimens in COVID-19.

The coronavirus disease 2019 (COVID-19) is an epidemic disease caused by the Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2) and spreading throughout the world rapidly. Here we evaluated the efficacy of the Lopinavir/Ritonavir (LPV/r) and its combination with other drugs in the treatment of COVID-19. We included 170 confirmed COVID-19 patients who had been cured and discharged. Their antiviral therapies were LPV/r alone or combinations with interferon (IFN), Novaferon and Arbidol. We evaluated the medication efficacy by comparing the time of the negative nucleic acid conversion and the length of hospitalization mainly. The LPV/r + Novaferon [6.00 (4.00-8.00) and 7.50 (5.00-10.00) days] had shorter time of the negative nucleic acid conversion (P = .0036) and shorter time of hospitalization (P < .001) compared with LPV/r alone [9.00 (5.00-12.00) and 12.00 (11.00-15.00) days] and LPV/r + IFN [9.00 (7.25-11.00) and 12.00 (10.00-13.50) days]. On the contrary, LPV/r + IFN [9.00 (7.25-11.00) and 12.00 (10.00-13.50) days] had shorter time of the negative nucleic acid conversion (P = .031) and shorter time of hospitalization (P < .001) compared with LPV/r + IFN +Novaferon [10.00 (8.00-11.25) and 13.50 (11.50-17.00) days] and LPV/r + IFN +Arbidol [14.00 (9.75-19.00) and 19.50 (13.25-24.00) days]. In conclusion, the combination of LPV/r and Novaferon may have better efficacy against COVID-19. However, adding IFN based on LPV/r + Novaferon or adding Arbidol based on LPV/r + IFN may not improve the efficacy.

Polymyxin B-Associated Nephrotoxicity and Its Predictors: A Retrospective Study in Carbapenem-Resistant Gram-Negative Bacterial Infections.

Polymyxin B (PMB), a kind of polymyxin, was widely used in carbapenem-resistant Gram-negative bacterial (CR-GNB) infections. However, adverse reactions such as nephrotoxicity and neurotoxicity limit its use in clinical practice. The aim of this study was to explore PMB associated with nephrotoxicity and its predictors. Patients who received PMB intravenous drip for more than 72 h were eligible for the study. Characteristics of patients, concomitant nephrotoxic agents, underlying disease, and antimicrobial susceptibility were submitted for descriptive analysis. Univariate analysis and binary logistic regression were used to assess the factors leading to acute kidney injury (AKI). AKI was assessed with serum creatinine variations according to the classification of risk (stage R), injury (stage I), failure (stage F), loss, and end-stage of kidney disease. Among 234 patients with CR-GNB infections who used PMB in our study, 67 (28.63%) patients developed AKI, including 31 (14.25%) patients in stage R, 15 (6.41%) patients in stage I, and 21 (8.97%) patients in stage F. The incident rate of PMB-related nephrotoxicity in patients with normal renal function was 32.82% (43/131). The higher risk factors of AKI include males [odds ratio (OR) = 3.237; 95% confidence interval (95%CI) = 1.426-7.350], digestive system diseases [OR = 2.481 (1.127-5.463)], using furosemide (>20 mg/day) [OR = 2.473 (1.102-5.551)], and baseline serum creatinine [OR = 0.994 (0.990-0.999)]. Nonparametric tests of K-independent samples showed that baseline serum creatinine and the PMB maintenance dose were associated with the severity of nephrotoxicity (both p < 0.05). Male, digestive system diseases, using furosemide (>20 mg/day), and high baseline serum creatinine were the independent risk factors of PMB-associated AKI development. The maintenance dose of PMB may be related to the severity of AKI. These risk factors should be taken into consideration when initiating PMB-based therapy. The serum creatinine value should be closely monitored when using PMB.

Abnormal Indexes of Liver and Kidney Injury Markers Predict Severity in COVID-19 Patients.

BACKGROUND: SARS-CoV-2 can damage not only the lungs but also the liver and kidney. Most critically ill patients with coronavirus disease 2019 (COVID-19) have liver and kidney dysfunction. We aim to investigate the levels of liver and kidney function indexes in mild and severe COVID-19 patients and their capability to predict the severity of the disease. METHODS: The characteristics and laboratory indexes were compared between patients with different conditions. We applied binary logistic regression to find the independent risk factors of severe patients. Receiver operating characteristic (ROC) analysis was used to predict the severity of COVID-19 using the liver and kidney function indexes. RESULTS: This study enrolled 266 COVID-19 patients, including 235 mild patients and 31 severe patients. Compared with mild patients, severe patients had lower albumin (ALB) and higher alanine aminotransferase (ALT), aspartate aminotransferase (AST), and urea nitrogen (BUN) (all p<0.001). Binary logistic regression analysis also identified ALB [OR=0.273 (0.079-0.947), p=0.041] and ALT [OR=2.680 (1.036-6.934), p=0.042] as independent factors of severe COVID-19 patients. Combining ALB, ALT, BUN, and LDH exhibited the area under ROC at 0.914, with a sensitivity of 86.7% and specificity of 83.0%. CONCLUSION: COVID-19 patients, especially severe patients, have damage to liver and kidney function. ALT, AST, LDH, and BUN could be independent factors for predicting the severity of COVID-19. Combining the ALB, ALT, BUN, and LDH could predict the transition from mild to severe in COVID-19 patients.

A Comparative Study of the Microbiological Efficacy of Polymyxin B on Different Carbapenem-Resistant Gram-Negative Bacteria Infections.

Objective: The emergence of carbapenem-resistant gram-negative bacteria (CR-GNB) has brought great challenges to clinical anti-infection treatment around the world. Polymyxins are often considered as the last line of defense in the treatment of CR-GNB infections. In this study, we explored the microbiological efficacy of Polymyxin B (PMB) on different CR-GNB infections as well as the factors influencing microbiological efficacy. Methods: CR-GNB infected patients with PMB-based regimens were enrolled. Clinical and microbiological data were collected from the medical electronic record system of the Second Xiangya hospital. The efficacy of PMB on different CR-GNB was evaluated by the clearance rate at 7-days and within the course of treatment, as well as the 30-day mortality rate. Results: A total of 294 CR-GNB infected patients were enrolled: 154 CR-Acinetobacter baumannii (CRAB), 55 CR-Klebsiella pneumoniae (CRKP), and 85 CR-Pseudomonas aeruginosa (CRPA). The CRAB group had the highest 7-day bacterial clearance rate [(CRAB: 39.0%) vs. (CRKP: 29.4%) vs. (CRPA: 14.5%), P = 0.003] and total bacterial clearance rate [(CRAB: 49.0%) vs. (CRKP: 39.8%) vs. (CRPA: 18.2%), P < 0.001] among the three groups, while the bacterial clearance rate of the CRPA group was the lowest. Multivariate logistic regression showed that the differences among the three groups were multiple CR-GNB infections (P = 0.004), respiratory infections (P = 0.001), PMB resistance (P < 0.001), and the combination of tigecycline (P < 0.001). Binary logistic regression showed that multiple CR-GNB infection [(7-day bacterial clearance: P = 0.004) & (total bacterial clearance: P = 0.011)] and bacterial species [(7-day bacterial clearance: P < 0.001) & (total bacterial clearance: P < 0.001)] were independent risk factors for microbiological efficacy. Conclusion: PMB exhibited differential microbiological efficacy on different types of CR-GNB infections; it had the best effect on CRAB, followed by CRKP and CRPA. Multiple CR-GNB infections and bacterial species were independent risk factors for microbiological efficacy.