BACKGROUND & AIMS:We performed a systematic review and meta-analysis to comprehensively estimate adenoma miss rate (AMR) and advanced AMR (AAMR) and explore associated factors. METHODS: We searched the PubMed, Web of Science, and Ovid EMBASE databases for studies published through April 2018 on tandem colonoscopies, with AMR and AAMR as the primary outcomes. We performed meta-regression analyses to identify risk factors and factors associated with outcome. Primary outcomes were AMR and AAMR and secondary outcomes were AMR and AAMR for different locations, sizes, pathologies, morphologies, and populations. RESULTS: In a meta-analysis of 43 publications and more than 15,000 tandem colonoscopies, we calculated miss rates of 26% for adenomas (95% confidence interval [CI] 23%-30%), 9% for advanced adenomas (95% CI 4%-16%), and 27% for serrated polyps (95% CI 16%-40%). Miss rates were high for proximal advanced adenomas (14%; 95% CI 5%-26%), serrated polyps (27%; 95% CI 16%-40%), flat adenomas (34%; 95% CI 24%-45%), and in patients at high risk for colorectal cancer (33%; 95% CI 26%-41%). Miss rates could be decreased by adequate bowel preparation and auxiliary techniques (P ¼ .06; P ¼ .04, and P ¼ .01, respectively). The adenoma detection rate (ADR), adenomas per index colonoscopy, and adenomas per positive index colonoscopy (APPC) were independently associated with AMR (P ¼ .02, P ¼ .01, and P ¼ .008, respectively), whereas APPC was the only factor independently associated with AAMR (P ¼ .006). An APPC value greater than 1.8 was more effective in monitoring AMR (31% vs 15% for AMR P < .0001) than an ADR value of at least 34% (27% vs 17% for AMR; P ¼ .008). The AAMR of colonoscopies with an APPC value below 1.7 was 35%, vs 2% for colonoscopies with an APPC value of at least 1.7 (P ¼ .0005). CONCLUSIONS: In a systematic review and meta-analysis, we found that adenomas and advanced adenomas are missed (based on AMR and AAMR) more frequently than previously believed. In addition to ADR,

No abstract

No abstract

With the rise of online e-commerce platforms, more and more customers prefer to shop online. To sell more products, online platforms introduce various modules to recommend items with different properties such as huge discounts. A web page often consists of different independent modules. The ranking policies of these modules are decided by different teams and optimized individually without cooperation, which might result in competition between modules. Thus, the global policy of the whole page could be sub-optimal. In this paper, we propose a novel multi-agent cooperative reinforcement learning approach with the restriction that different modules cannot communicate. Our contributions are threefold. Firstly, inspired by a solution concept in game theory named correlated equilibrium, we design a signal network to promote cooperation of all modules by generating signals (vectors) for different modules. Secondly, an entropy-regularized version of the signal network is proposed to coordinate agentsâȂŹ exploration of the optimal global policy. Furthermore, experiments based on real-world e-commerce data demonstrate that our algorithm obtains superior performance over baselines. CCS CONCEPTS • Information systems → Recommender systems; • Computing methodologies → Multi-agent systems.

Modified-masked T-2 toxin (mT-2) formed during metabolism in edible aquatic animals may go undetected by traditional analytical methods, thereby underestimating T-2 toxicity. The effects of T-2 on growth and antioxidant capacity and histopathological changes in the hepatopancreas were studied in Pacific white shrimp Litopenaeus vannamei exposed for 20 d to 0, 0.5, 1.2, 2.4, 4.8, and 12.2 mg/kg of T-2 in their feed. The concentration of mT-2 in the hepatopancreas was detected by liquid chromatography-tandem mass spectrophotometry before and after trifluoroacetic acid (TFA) treatment that converted mT-2 to free T-2. A dose-dependent increase in mT-2 concentration was observed in the hepatopancreas. Dietary exposure to T-2 significantly decreased (P < 0.05) shrimp growth and survival rate compared with the controls. The malondialdehyde (MDA) concentration was significantly increased in shrimp exposed to feed with ≥2.4 mg/kg T-2 (P < 0.05). The antioxidant enzymes, superoxide dismutase (SOD) and glutathione peroxidase (GPx), total antioxidant capacity (T-AOC), and also glutathione (GSH) content increased in shrimp dosed with 2.4-4.8 mg/kg T-2 but declined at the highest dose (12.2 mg/kg), probably indicating an inability to cope with high concentrations of reactive oxygen species (ROS) as evident from a marked increase in MDA (P < 0.05) culminating in cellular toxicity. Histopathological changes in the hepatopancreas were dose dependent, with cell autophagy evident at the highest exposure dose. This is the first report in shrimp of a dose-dependent increase in ROS, SOD enzyme activity, and T-AOC at low T-2 exposures, and associated histopathological changes in the hepatopancreas, in response to dietary T-2. Received January 26, 2016; accepted October 9, 2016.

Vibrio parahaemolyticus is a seafood opportunistic pathogen. There are evidences suggesting that virulence skills, including hemolytic activity and biofilm formation, are regulated by the luxM/luxS-dependent quorum-sensing system in V. parahaemolyticus, and their regulatory mechanism is not well understood. To better understand the virulence regulatory mechanism of V. parahaemolyticus, the luxM deletion (△luxM) and luxS deletion (△luxS) mutants were constructed and their impacts on growth, hemolysin activity, and biofilm were investigated. Results show that both luxM and luxS are involved in the adaptation to environmental conditions in early adaptive-log phase growth of V. parahaemolyticus. Thermostable direct hemolysin gene (tdh) was negatively regulated by luxM and positively regulated by luxS. The biofilm formation was negatively regulated by both luxS and luxM. This study provides an insight into some aspects of V. parahaemolyticus virulence regulation by luxM/luxS-dependent quorum-sensing system.

Rationale, aims, and objectives: Acute postoperative pain can result in immune dysfunction, which can be partly mitigated by efficient pain management. Opioids that have been widely applied to analgesia have been shown to suppress immune function, which has a negative impact on the treatment of patients with cancer. This study investigated the effects of perioperative fentanyl analgesia alone or in combination with parecoxib sodium on postoperative pain, immune function, and prognosis in patients undergoing hepatectomy of hepatocellular carcinoma (HCC).Methods: A total of 80 patients scheduled for hepatectomy between October 2013 and August 2014 were included. Patients were randomly divided into two groups (n = 40) and allocated to receive parecoxibsodium 40 mg (group P) or placebo (group C) 30 minutes before induction of anaesthesia, followed by 40 mg every 12 hours for 48 hours after the operation. All patients had access to patient-controlled analgesia with intravenous fentanylpostoperatively. Venous blood samples were collected at the following time points: 30 minutes before induction of anaesthesia (T0), the end of the surgery (T1), 24 hours after surgery (T2), and 72 hours after surgery (T3). The percentages of CD3+, CD4+, CD8+, CD4+/CD8+ T cells, and CD3+CD16+CD56+ (NK) cells at these time points were quantified by flow cytometry (FCM).Visual analogue scale (VAS) scores, total fentanyl consumption, and adverse effects were recorded. The prognostic differences in overall survival (OS) and diseasefree survival (DFS) between the two groups was also investigated.Results: For both groups, the percentages of CD3+, CD4+ T cells, and the ratio of CD4+/CD8+ significantly decreased at T1 and T2 (P < .05). The percentages of CD3+ T cells were significantly lower in group C than that in group P at T2 (P < .05). In group C, the amount of CD3+ T cells was lower at T3 compared with T0 (P < .05).The percentages of NK cells significantly decreased at T1 in both groups (P < .05).

Hierarchical reinforcement learning (HRL) is a promising approach to solve tasks with long time horizons and sparse rewards. It is often implemented as a high-level policy assigning subgoals to a low-level policy. However, it suffers the high-level non-stationarity problem since the low-level policy is constantly changing. The non-stationarity also leads to the data efficiency problem: policies need more data at non-stationary states to stabilize training. To address these issues, we propose a novel HRL method: Interactive Influence-based Hierarchical Reinforcement Learning (I^2HRL). First, inspired by agent modeling, we enable the interaction between the low-level and high-level policies to stabilize the high-level policy training. The high-level policy makes decisions conditioned on the received low-level policy representation as well as the state of the environment. Second, we furthermore stabilize the high-level policy via an information-theoretic regularization with minimal dependence on the changing low-level policy. Third, we propose the influence-based exploration to more frequently visit the non-stationary states where more transition data is needed. We experimentally validate the effectiveness of the proposed solution in several tasks in MuJoCo domains by demonstrating that our approach can significantly boost the learning performance and accelerate learning compared with state-of-the-art HRL methods.