Effective attention-based network for syndrome differentiation of AIDS.

BACKGROUND: Syndrome differentiation aims at dividing patients into several types according to their clinical symptoms and signs, which is essential for traditional Chinese medicine (TCM). Several previous works were devoted to employing the classical algorithms to classify the syndrome and achieved delightful results. However, the presence of ambiguous symptoms substantially disturbed the performance of syndrome differentiation, This disturbance is always due to the diversity and complexity of the patients' symptoms. METHODS: To alleviate this issue, we proposed an algorithm based on the multilayer perceptron model with an attention mechanism (ATT-MLP). In particular, we first introduced an attention mechanism to assign different weights for different symptoms among the symptomatic features. In this manner, the symptoms of major significance were highlighted and ambiguous symptoms were restrained. Subsequently, those weighted features were further fed into an MLP to predict the syndrome type of AIDS. RESULTS: Experimental results for a real-world AIDS dataset show that our framework achieves significant and consistent improvements compared to other methods. Besides, our model can also capture the key symptoms corresponding to each type of syndrome. CONCLUSION: In conclusion, our proposed method can learn these intrinsic correlations between symptoms and types of syndromes. Our model is able to learn the core cluster of symptoms for each type of syndrome from limited data, while assisting medical doctors to diagnose patients efficiently.

Graph Representation Learning Based on Specific Subgraphs for Biomedical Interaction Prediction.

Discovering the novel associations of biomedical entities is of great significance and can facilitate not only the identification of network biomarkers of disease but also the search for putative drug targets. Graph representation learning (GRL) has incredible potential to efficiently predict the interactions from biomedical networks by modeling the robust representation for each node. However, the current GRL-based methods learn the representation of nodes by aggregating the features of their neighbors with equal weights. Furthermore, they also fail to identify which features of higher-order neighbors are integrated into the representation of the central node. In this work, we propose a novel graph representation learning framework: a multi-order graph neural network based on reconstructed specific subgraphs (MGRS) for biomedical interaction prediction. In the MGRS, we apply the multi-order graph aggregation module (MOGA) to learn the wide-view representation by integrating the multi-hop neighbor features. Besides, we propose a subgraph selection module (SGSM) to reconstruct the specific subgraph with adaptive edge weights for each node. SGSM can clearly explore the dependency of the node representation on the neighbor features and learn the subgraph-based representation based on the reconstructed weighted subgraphs. Extensive experimental results on four public biomedical networks demonstrate that the MGRS performs better and is more robust than the latest baselines.

A network medicine-based approach to explore the relationship between depression and inflammation.

Background: Depression is widespread global problem that not only severely impacts individuals' physical and mental health but also imposes a heavy disease burden on nations and societies. The role of inflammation in the pathogenesis and pathophysiology of depression has received much attention, but the precise relationship between the two remains unclear. This study aims to investigate the correlation between depression and inflammation using a network medicine approach. Methods: We utilized a degree-preserving approach to identify the large connected component (LCC) of all depression-related proteins in the human interactome. The LCC was deemed as the disease module for depression. To measure the association between depression and other diseases, we calculated the overlap between these disease protein modules using the Sab algorithm. A smaller Sab value indicates a stronger association between diseases. Building on the results of this analysis, we further explored the correlation between inflammation and depression by conducting enrichment and pathway analyses of critical targets. Finally, we used a network proximity approach to calculate drug-disease proximity to predict the efficacy of drugs for the treatment of depression. We calculated and ranked the distances between depression disease modules and 6,100 drugs. The top-ranked drugs were selected to explore their potential for treating depression based on the hypothesis that their antidepressant effects are related to reducing inflammation. Results: In the human interactome, all depression-related proteins are clustered into a large connected component (LCC) consisting of 202 proteins and multiple small subgraphs. This indicates that depression-related proteins tend to form clusters within the same network. We used the 202 LCC proteins as the key disease module for depression. Next, we investigated the potential relationships between depression and 299 other diseases. Our analysis identified over 18 diseases that exhibited significant overlap with the depression module. Where SAB = -0.075 for the vascular disease and depressive disorders module, SAB = -0.070 for the gastrointestinal disease and depressive disorders module, and SAB = -0.062 for the endocrine system disease and depressive disorders module. The distance between them SAB < 0 implies that the pathogenesis of depression is likely to be related to the pathogenesis of its co-morbidities of depression and that potential therapeutic approaches may be derived from the disease treatment libraries of these co-morbidities. Further, considering that the inflammation is ubiquitous in some disease, we calculate the overlap between the collected inflammation module (236 proteins) and the depression module (202 proteins), finding that they are closely related (Sdi = -0.358) in the human protein interaction network. After enrichment and pathway analysis of key genes, we identified the HIF-1 signaling pathway, PI3K-Akt signaling pathway, Th17 cell differentiation, hepatitis B, and inflammatory bowel disease as key to the inflammatory response in depression. Finally, we calculated the Z-score to determine the proximity of 6,100 drugs to the depression disease module. Among the top three drugs identified by drug-disease proximity analysis were Perphenazine, Clomipramine, and Amitriptyline, all of which had a greater number of targets in the network associated with the depression disease module. Notably, these drugs have been shown to exert both anti-inflammatory and antidepressant effects, suggesting that they may modulate depression through an anti-inflammatory mechanism. These findings demonstrate a correlation between depression and inflammation at the network medicine level, which has important implications for future elucidation of the etiology of depression and improved treatment outcomes. Conclusion: Neuroimmune signaling pathways play an important role in the pathogenesis of depression, and many classes of antidepressants exhibiting anti-inflammatory properties. The pathogenesis of depression is closely related to inflammation.

Composed Image Retrieval via Explicit Erasure and Replenishment With Semantic Alignment.

Composed image retrieval aims at retrieving the desired images, given a reference image and a text piece. To handle this task, two important subprocesses should be modeled reasonably. One is to erase irrelated details of the reference image against the text piece, and the other is to replenish the desired details in the image against the text piece. Nowadays, the existing methods neglect to distinguish between the two subprocesses and implicitly put them together to solve the composed image retrieval task. To explicitly and orderly model the two subprocesses of the task, we propose a novel composed image retrieval method which contains three key components, i.e., Multi-semantic Dynamic Suppression module (MDS), Text-semantic Complementary Selection module (TCS), and Semantic Space Alignment constraints (SSA). Concretely, MDS is to erase irrelated details of the reference image by suppressing its semantic features. TCS aims to select and enhance the semantic features of the text piece and then replenish them to the reference image. In the end, to facilitate the erasure and replenishment subprocesses, SSA aligns the semantics of the two modality features in the final space. Extensive experiments on three benchmark datasets (Shoes, FashionIQ, and Fashion200K) show the superior performance of our approach against state-of-the-art methods.