SHAPE: A Sample-Adaptive Hierarchical Prediction Network for Medication Recommendation.

Effectively medication recommendation with complex multimorbidity conditions is a critical yet challenging task in healthcare. Most existing works predicted medications based on longitudinal records, which assumed the encoding format of intra-visit medical events are serialized and information transmitted patterns of learning longitudinal sequence data are stable. However, the following conditions may have been ignored: 1) A more compact encoder for intra-relationship in the intra-visit medical event is urgent; 2) Strategies for learning accurate representations of the variable longitudinal sequences of patients are different. In this article, we proposed a novel Sample-adaptive Hierarchical medicAtion Prediction nEtwork, termed SHAPE, to tackle the above challenges in the medication recommendation task. Specifically, we design a compact intra-visit set encoder to encode the relationship in the medical event for obtaining visit-level representation and then develop an inter-visit longitudinal encoder to learn the patient-level longitudinal representation efficiently. To endow the model with the capability of modeling the variable visit length, we introduce a soft curriculum learning method to assign the difficulty of each sample automatically by the visit length. Extensive experiments on a benchmark dataset verify the superiority of our model compared with several state-of-the-art baselines.

Mechanism of activation and biased signaling in complement receptor C5aR1.

The complement system plays an important role in the innate immune response to invading pathogens. The complement fragment C5a is one of its important effector components and exerts diverse physiological functions through activation of the C5a receptor 1 (C5aR1) and associated downstream G protein and ß-arrestin signaling pathways. Dysfunction of the C5a-C5aR1 axis is linked to numerous inflammatory and immune-mediated diseases, but the structural basis for activation and biased signaling of C5aR1 remains elusive. Here, we present cryo-electron microscopy structures of the activated wild-type C5aR1-Gi protein complex bound to each of the following: C5a, the hexapeptidic agonist C5apep, and the G protein-biased agonist BM213. The structures reveal the landscape of the C5a-C5aR1 interaction as well as a common motif for the recognition of diverse orthosteric ligands. Moreover, combined with mutagenesis studies and cell-based pharmacological assays, we deciphered a framework for biased signaling using different peptide analogs and provided insight into the activation mechanism of C5aR1 by solving the structure of C5aR1I116A mutant-Gi signaling activation complex induced by C089, which exerts antagonism on wild-type C5aR1. In addition, unusual conformational changes in the intracellular end of transmembrane domain 7 and helix 8 upon agonist binding suggest a differential signal transduction process. Collectively, our study provides mechanistic understanding into the ligand recognition, biased signaling modulation, activation, and Gi protein coupling of C5aR1, which may facilitate the future design of therapeutic agents.

Multimodal Data Matters: Language Model Pre-Training Over Structured and Unstructured Electronic Health Records.

As two important textual modalities in electronic health records (EHR), both structured data (clinical codes) and unstructured data (clinical narratives) have recently been increasingly applied to the healthcare domain. Most existing EHR-oriented studies, however, either focus on a particular modality or integrate data from different modalities in a straightforward manner, which usually treats structured and unstructured data as two independent sources of information about patient admission and ignore the intrinsic interactions between them. In fact, the two modalities are documented during the same encounter where structured data inform the documentation of unstructured data and vice versa. In this paper, we proposed a Medical Multimodal Pre-trained Language Model, named MedM-PLM, to learn enhanced EHR representations over structured and unstructured data and explore the interaction of two modalities. In MedM-PLM, two Transformer-based neural network components are firstly adopted to learn representative characteristics from each modality. A cross-modal module is then introduced to model their interactions. We pre-trained MedM-PLM on the MIMIC-III dataset and verified the effectiveness of the model on three downstream clinical tasks, i.e., medication recommendation, 30-day readmission prediction and ICD coding. Extensive experiments demonstrate the power of MedM-PLM compared with state-of-the-art methods. Further analyses and visualizations show the robustness of our model, which could potentially provide more comprehensive interpretations for clinical decision-making.

CATNet: Cross-event attention-based time-aware network for medical event prediction.

Medical event prediction (MEP) is a fundamental task in the healthcare domain, which needs to predict medical events, including medications, diagnosis codes, laboratory tests, procedures, outcomes, and so on, according to historical medical records of patients. Many researchers have tried to build MEP models to overcome the challenges caused by the heterogeneous and irregular temporal characteristics of EHR data. However, most of them consider the heterogenous and temporal medical events separately and ignore the correlations among different types of medical events, especially relations between heterogeneous historical medical events and target medical events. In this paper, we propose a novel neural network based on attention mechanism called Cross-event Attention-based Time-aware Network (CATNet) for MEP. It is a time-aware, event-aware and task-adaptive method with the following advantages: 1) modeling heterogeneous information and temporal information in a unified way and considering irregular temporal characteristics locally and globally respectively, 2) taking full advantage of correlations among different types of events via cross-event attention. Experiments on two public datasets (MIMIC-III and eICU) show CATNet outperforms other state-of-the-art methods on various MEP tasks. The source code of CATNet is released at https://github.com/sherry6247/CATNet.git.

Multi-channel fusion LSTM for medical event prediction using EHRs.

Automatic medical event prediction (MEP), e.g. diagnosis prediction, medication prediction, using electronic health records (EHRs) is a popular research direction in health informatics. In many cases, MEP relies on the determinations from different types of medical events, which demonstrates the heterogeneous nature of EHRs. However, most existing methods for MEP fail to distinguishingly model the type of event that is highly associated with the prediction task, i.e. task-wise event, which usually plays a more significant role than other events. In this paper, we proposed a Long Short-Term Memory network (LSTM)-based method for MEP, named Multi-Channel Fusion LSTM (MCF-LSTM), which models the correlations between different types of medical events using multiple network channels. To this end, we designed a task-wise fusion module, in which a gated network is applied to select how much information can be transferred between events. Furthermore, the irregular temporal interval between adjacent medical visits is also modeled in an individual channel, which is combined with other events in a unified manner. We compared MCF-LSTM with state-of-the-art methods on four MEP tasks on two public datasets: MIMIC-III and eICU. Experimental results show that MCF-LSTM achieves promising results on AUC(receiver operating characteristic curve), AUPR (area under the precision-recall curve), and top-k recall, and outperforms other methods with high stability.

Comparison of improving energy use and mitigating pollutant emissions from industrial and non-industrial activities: Evidence from a variable-specific productivity analysis framework.

This paper seeks to account for differences in productivity of the industrial and non-industrial activities in the productivity analysis framework. The Luenberger productivity indicator is widely applied to analyze the productivity change, and can be decomposed as it follows the additive structure. But there have been few studies on sector operation performance and industrial structure involving both the industrial and non-industrial inputs, output and air pollutant emissions. Resorting on the China's province-level data on energy, output and air pollutants from 2006 to 2019, we find that the industrial SO2 emissions, energy consumption and NOX emissions are the major factors leading to sector operation inefficiency. By decomposing the operation performance indicator (OPI), we observe that contribution to productivity change by energy consumption, air pollutant emissions and output is higher than contribution by the non-industrial variables. Furthermore, technical progress offsets negative efficiency growth. In order to implement energy conservation, emissions reduction and industrial restructuring at the provincial level, China's government should take efforts to improve the efficiency of non-industrial energy consumption and support the development of cleaner industries.

Large-Scale Spraying of Roads with Water Contributes to, Rather Than Prevents, Air Pollution.

Spraying roads with water on a large scale in Chinese cities is one of the supplementary precaution or mitigation actions implemented to control severe air pollution events or heavy haze-fog events in which the mechanisms causing them are not yet fully understood. These air pollution events were usually characterized by higher air humidity. Therefore, there may be a link between this action and air pollution. In the present study, the impact of water spraying on the PM2.5 concentration and humidity in air was assessed by measuring chemical composition of the water, undertaking a simulated water spraying experiment, measuring residues and analyzing relevant data. We discovered that spraying large quantities of tap or river water on the roads leads to increased PM2.5 concentration and humidity, and that daily continuous spraying produces a cumulative effect on air pollution. Spraying the same amount of water produces greater increases in humidity and PM2.5 concentration during cool autumn and winter than during hot summer. Our results demonstrate that spraying roads with water increases, rather than decreases, the concentration of PM2.5 and thus is a new source of anthropogenic aerosol and air pollution. The higher vapor content and resultant humidity most likely create unfavorable meteorological conditions for the dispersion of air pollution in autumn and winter with low temperature.

B7-H4 is a potential prognostic biomarker of prostate cancer.

B7-H4 is a member of B7 family which regulates immune responses by delivering costimulatory signals. However, it negatively regulates T cell-mediated immunity and may play an important role in tumor immune evasion. Although several studies have been reported that expression of B7-H4 is elevated in the several types of human cancer with a poor clinical outcome, its clinical significance in the prostate cancer (PCa) has not been well studied. In this study, we investigated the clinical significance of B7-H4 in human PCa and determined if B7-H4 expression is associated with the cancer cell stemness in PCa. Our studies show that expression of B7-H4 is correlated with the pathologic tumor (pT) stage and the clinical stage of PCa. The Kaplan-Meier survival analysis revealed that PCa patients with high expression of B7-H4 exhibits a shorter overall survival (OS) rate. Univariate and multivariate Cox regression analysis indicated that B7-H4 is an independent poor prognostic factor of PCa. In addition, the expression of B7-H4 is correlated with the cancer cell stemness associated genes expression in PCa. Further, our studies show that B7-H4 regulates cancer cell stemness associated genes expression and effects on the cell cycle and PI3K/Akt signaling related genes expression in PCa. These results indicate that B7-H4 expression is associated with cancer cell stemness, and B7-H4 is a potential prognostic biomarker and a therapeutic target of PCa.

A hybrid method of recurrent neural network and graph neural network for next-period prescription prediction.

Electronic health records (EHRs) have been widely used to help physicians to make decisions by predicting medical events such as diseases, prescriptions, outcomes, and so on. How to represent patient longitudinal medical data is the key to making these predictions. Recurrent neural network (RNN) is a popular model for patient longitudinal medical data representation from the view of patient status sequences, but it cannot represent complex interactions among different types of medical information, i.e., temporal medical event graphs, which can be represented by graph neural network (GNN). In this paper, we propose a hybrid method of RNN and GNN, called RGNN, for next-period prescription prediction from two views, where RNN is used to represent patient status sequences, and GNN is used to represent temporal medical event graphs. Experiments conducted on the public MIMIC-III ICU data show that the proposed method is effective for next-period prescription prediction, and RNN and GNN are mutually complementary.

Air pollutants and outpatient visits for cardiovascular disease in a severe haze-fog city: Shijiazhuang, China.

BACKGROUND: Many studies have reported the impact of air pollution on cardiovascular disease (CVD), but few of these studies were conducted in severe haze-fog areas. The present study focuses on the impact of different air pollutant concentrations on daily CVD outpatient visits in a severe haze-fog city. METHODS: Data regarding daily air pollutants and outpatient visits for CVD in 2013 were collected, and the association between six pollutants and CVD outpatient visits was explored using the least squares mean (LSmeans) and logistic regression. Adjustments were made for days of the week, months, air temperature and relative humidity. RESULTS: The daily CVD outpatient visits for particulate matter (PM10 and PM2.5), sulphur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), and ozone (O3) in the 90th-quantile group were increased by 30.01, 29.42, 17.68, 14.98, 29.34%, and - 19.87%, respectively, compared to those in the <10th-quantile group. Odds ratios (ORs) and 95% confidence intervals (CIs) for the increase in daily CVD outpatient visits in PM10 300- and 500-µg/m3, PM2.5 100- and 300-µg/m3 and CO 3-mg/m3 groups were 2.538 (1.070-6.020), 7.781 (1.681-36.024), 3.298 (1.559-6.976), 8.72 (1.523-49.934), and 5.808 (1.016-33.217), respectively, and their corresponding attributable risk percentages (AR%) were 60.6, 87.15, 69.68, 88.53 and 82.78%, respectively. The strongest associations for PM10, PM2.5 and CO were found only in lag 0 and lag 1. The ORs for the increase in CVD outpatient visits per increase in different units of the six pollutants were also analysed. CONCLUSIONS: All five air pollutants except O3 were positively associated with the increase in daily CVD outpatient visits in lag 0. The high concentrations of PM10, PM2.5 and CO heightened not only the percentage but also the risk of increased daily CVD outpatient visits. PM10, PM2.5 and CO may be the main factors of CVD outpatient visits.

Target silencing of components of the conserved oligomeric Golgi complex impairs HIV-1 replication.

All viruses require host cell factors to replicate. A large number of host factors have been identified that participate at numerous points of the human immunodeficiency virus 1 (HIV-1) life cycle. Recent evidence supports a role for components of the trans-Golgi network (TGN) in mediating early steps in the HIV-1 life cycle. The conserved oligomeric Golgi (COG) complex is a heteroctamer complex that functions in coat protein complex I (COPI)-mediated intra-Golgi retrograde trafficking and plays an important role in the maintenance of Golgi structure and integrity as well as glycosylation enzyme homeostasis. The targeted silencing of components of lobe B of the COG complex, namely COG5, COG6, COG7 and COG8, inhibited HIV-1 replication. This inhibition of HIV-1 replication preceded late reverse transcription (RT) but did not affect viral fusion. Silencing of the COG interacting protein the t-SNARE syntaxin 5, showed a similar defect in late RT product formation, strengthening the role of the TGN in HIV replication.