Multimodal affective computing, learning to recognize and interpret human affect and subjective information from multiple data sources, is still challenging because: (i) it is hard to extract informative features to represent human affects from heterogeneous inputs; (ii) current fusion strategies only fuse different modalities at abstract levels, ignoring time-dependent interactions between modalities. Addressing such issues, we introduce a hierarchical multimodal architecture with attention and word-level fusion to classify utterancelevel sentiment and emotion from text and audio data. Our introduced model outperforms state-of-the-art approaches on published datasets, and we demonstrate that our model's synchronized attention over modalities offers visual interpretability.

Data on the association of dietary patterns with non-alcoholic fatty liver disease (NAFLD) among adolescents are scarce. Hence, the purpose of this study was to ascertain the influence of dietary patterns and key foods on NAFLD among adolescents in Shandong, China. Data were extracted from Linyi Nutrition and Health study during 2015 to 2016. This cross-sectional study population comprised 1639 participants aged between 16 and 23 years. Dietary intake was assessed by the use of a semiquantitative food frequency questionnaire (FFQ), containing 85 food items. NAFLD diagnosis was defined as individuals whose ultrasound examination disclosed hepatic steatosis at any stage, in the absence of excess intake of alcoholic beverages. The odds ratio (OR) and 95% confidence interval (CI) were estimated for each quartile of the dietary pattern adherence scores using logistic regression analysis. Of 1639 participants, 221 (13.5%) were classified as having NAFLD. Three major dietary patterns were derived from factor analysis: traditional Chinese, Western, and high-energy dietary patterns. There were significant differences in the intake of whole grains, tuber, and vegetable across quartiles of the traditional Chinese and Western pattern (P < .05). Besides, compared with adolescents in the lowest quartile, those in the highest quartile for whole grains intake had a lower OR for NAFLD (OR = 0.72; 95%CI: 0.61–0.98; P < .05), and for red meat and soft drink consumption had greater OR for NAFLD (OR = 1.34; 95% CI: 1.06–1.72; OR = 1.69; 95% CI: 1.34–2.56; respectively, P < .05). After adjustment for several potential confounders, participants in the highest quartile of the traditional Chinese pattern scores had lower OR for NAFLD (OR = 0.726; 95% CI: 0.383–0.960, P < .05) than did those in the lowest quartile, whereas those in the highest quartile of the Western pattern score had greater OR for NAFLD (OR = 1.197; 95% CI: 1.013–1.736, P < .01) than did those in the lowest quartile. No statistically significant association was found between the high-energy pattern and the risk of NAFLD.Our findings demonstrated that the traditional Chinese dietary pattern was associated with a lower risk, whereas the Western dietary pattern was associated with a higher risk of NAFLD.

No abstract