FASTNN: A Deep Learning Approach for Traffic Flow Prediction Considering Spatiotemporal Features.

Traffic flow forecasting is a critical input to intelligent transportation systems. Accurate traffic flow forecasting can provide an effective reference for implementing traffic management strategies, developing travel route planning, and public transportation risk assessment. Recent deep learning approaches of spatiotemporal neural networks to predict traffic flow show promise, but could be difficult to separately model the spatiotemporal aggregation in traffic data and intrinsic correlation or redundancy of spatiotemporal features extracted by the filter of the convolutional network. This can introduce biases in the predictions that interfere with subsequent planning decisions in transportation. To solve the mentioned problem, the filter attention-based spatiotemporal neural network (FASTNN) was proposed in this paper. First, the model used 3-dimensional convolutional neural networks to extract universal spatiotemporal dependencies from three types of historical traffic flow, the residual units were employed to prevent network degradation. Then, the filter spatial attention module was constructed to quantify the spatiotemporal aggregation of the features, thus enabling dynamic adjustment of the spatial weights. To model the intrinsic correlation and redundancy of features, this paper also constructed a lightweight module, named matrix factorization based resample module, which automatically learned the intrinsic correlation of the same features to enhance the concentration of the model on information-rich features, and used matrix factorization to reduce the redundant information between different features. The FASTNN has experimented on two large-scale real datasets (TaxiBJ and BikeNYC), and the experimental results show that the FASTNN has better prediction performance than various baselines and variant models.

Epidemiological, clinical and histological characteristics of HBV/HDV co-infection: a retrospective cross-sectional study in Guangdong, China.

BACKGROUND: The epidemiology of hepatitis D virus (HDV) in China is fairly unknown. The mechanisms whereby HDV leads to accelerated liver disease in hepatitis B virus (HBV)/HDV co-infected patients and the histological characteristics of chronic hepatitis D (CHD) patients need further investigation. METHODS: The prevalence of HDV was retrospectively evaluated in all consecutive hospitalized patients with chronic HBV infection from May 2005 to October 2011. HBV/HDV co-infected patients and HBV mono-infected patients were compared clinically and histologically. Significant histological abnormality was defined as significant necroinflammation (grade ≥A2) and/or significant fibrosis (stage ≥ F2). RESULTS: 6.5% of patients (426/6604) tested positive for IgM anti-HDV. HDV was more common in patients over 50 years old than those under 50 (11.7% vs. 5.1%, P<0.001). HBV/HDV co-infected patients had higher frequencies of end-stage liver disease (ESLD) than HBV mono-infected patients, and HDV co-infection was an independent risk factor for ESLD (OR: 1.428, 95%CI: 1.116-1.827; P = 0.005). The HBV DNA levels in the HBV/HDV group were significantly lower than the HBV group in chronic hepatitis patients (median: 6.50 log10copies/mL vs 6.80 log10copies/mL, P = 0.003), but higher than the HBV group in ESLD patients (median: 5.73 log10copies/mL vs 5.16 log10copies/mL, P<0.001). When stratified by alanine aminotransferase (ALT) level, 46.7%, 56.5% and 80.5% of CHD patients had significant necroinflammation and 86.7%, 87.0% and 90.3% had significant fibrosis with ALT 1-2×upper limit normal (ULN), 2-5×ULN and>5×ULN respectively. CONCLUSION: The prevalence of HDV is not low in patients with chronic HBV infection. HDV may contribute to progression to ESLD through late-phase HBV DNA reactivation.

Significant fibrosis is not rare in Chinese chronic hepatitis B patients with persistent normal ALT.

BACKGROUND: Limited studies have been done on chronic hepatitis B (CHB) patients defined according to the latest Asian-Pacific Association for the Study of the Liver guideline with liver histology by a large sample size. METHODS: We retrospectively evaluated liver histological characteristics on a cohort of consecutive treatment-naive CHB patients with persistent normal alanine aminotransferase (PNALT) or elevated ALT from May 2005 to October 2011. Histological assessment was based on the Metavir scoring system, significant abnormality was defined as necroinflammation grade ≥A2 and/or fibrosis stage ≥F2. RESULTS: A total of 675 CHB patients were recruited, including 516 HBeAg-positive and 159 HBeAg-negative patients. In HBeAg-positive patients, significant fibrosis was found 49.4% (42/85) in PNALT, 69.8% (88/126) in ALT 1-2×upper limit normal (ULN) and 81.6% (249/305) in ALT>2×ULN group, respectively. In HBeAg-negative patients, significant fibrosis was found 30.9% (17/55) in PNALT, 73.3% (33/45) in ALT 1-2×ULN and 94.9% (56/59) in ALT>2×ULN group, respectively. HBeAg-positive patients with PNALT over 30 years old had a higher frequency of significant fibrosis than those under 30 years old (87.5% vs. 45.5%, P = 0.058). Multivariate logistic regression analysis indicated increasing age (P = 0.012), higher aspartate aminotransferase (AST) (P < 0.001) and lower HBV DNA (P < 0.001) were associated with significant necroinflammation, while higher AST (P < 0.001), lower albumin (P = 0.027) and HBV DNA (P = 0.004) were associated with significant fibrosis in HBeAg-positive patients with elevated ALT. Higher AST was associated with significant necroinflammation in HBeAg-negative patients with elevated ALT (P = 0.009). CONCLUSIONS: Significant fibrosis is not rare in Chinese CHB patients with PNALT, especially HBeAg-positive patients over 30 years old.