Search

Abstract Background The identification of drug side effects plays a critical role in drug repositioning and drug screening. While clinical experiments yield accurate and reliable information about drug-related side effects, they are costly and time-consuming. Computational models have emerged as a promising alternative to predict the frequency of drug-side effects. However, earlier research has primarily centered on extracting and utilizing representations of drugs, like molecular structure or interaction graphs, often neglecting the inherent biomedical semantics of drugs and side effects. Results To address the previously mentioned issue, we introduce a hybrid multi-modal fusion framework (HMMF) for predicting drug side effect frequencies. Considering the wealth of biological and chemical semantic information related to drugs and side effects, incorporating multi-modal information offers additional, complementary semantics. HMMF utilizes various encoders to understand molecular structures, biomedical textual representations, and attribute similarities of both drugs and side effects. It then models drug-side effect interactions using both coarse and fine-grained fusion strategies, effectively integrating these multi-modal features. Conclusions HMMF exhibits the ability to successfully detect previously unrecognized potential side effects, demonstrating superior performance over existing state-of-the-art methods across various evaluation metrics, including root mean squared error and area under receiver operating characteristic curve, and shows remarkable performance in cold-start scenarios.

Abstract Background The efficacy and validity of excimer laser ablation (ELA) in the in-stent restenosis (ISR) has been confirmed. However, its application in de novo atherosclerotic lesions of lower extremity artery disease (LEAD) has not been clearly defined and its procedure has not been standardized. Methods ELABORATE is a prospective, multicenter, real-world study designed to evaluate the efficacy and safety between ELA combined with drug-coated balloon (DCB) and DCB alone in de novo atherosclerotic lesions of LEAD. Discussion ELABORATE is a prospective, multicenter, real-world study designed to assess the efficacy and safety between ELA combined with drug-coated balloon (DCB) and DCB alone in patients with de novo atherosclerotic lesions of LEAD. According to the real-world situation, eligible patients will be allocated to ELA + DCB group (group E) and DCB group (group C). Baseline and follow-up information (at 3, 6, and 12 months) will be collected. The primary efficacy point is primary patency at 12-months, and the secondary efficacy points include clinically driven target lesion reintervention (CD-TLR), change of Rutherford class, ankle-brachial index and ulcer healing rate. These indexes will be assessed and recorded at 3, 6, and 12-month follow-up. Also, safety evaluation, including major adverse event, all-cause mortality through 30-day follow-up, unplanned major amputation, bailout stent and distal embolization, will also be evaluated by an independent core laboratory. All the data will be collected and recorded by the electric data capture system. This study will be finished in 3 years and the 12-month results will be available in 2023. All the patients will be followed for 5 years. Trial registration number Chinese Clinical Trial Registry (ChiCTR2100051263). Registered 17 September 2019. http://www.chictr.org.cn/listbycreater.aspx .

Most existing methods for sorting synthetic aperture radar (SAR) emitter signals rely on either unsupervised clustering or supervised classification methods. However, unsupervised clustering can consume a significant amount of computational and storage space and is sensitive to the setting of hyperparameters, while supervised classification requires a considerable number of labeled samples. To address these limitations, we propose a self-supervised clustering-based method for sorting SAR radiation source signals. The method uses a constructed affinity propagation-convolutional neural network (AP-CNN) to perform self-supervised clustering of a large number of unlabeled signal time-frequency images into multiple clusters in the first stage. Subsequently, it uses a self-organizing map (SOM) network combined with inter-pulse parameters for further sorting in the second stage. The simulation results demonstrate that the proposed method outperforms other depth models and conventional methods in the environment where Gaussian white noise affects the signal. The experiments conducted using measured data also show the superiority of the proposed method in this paper.