Evaluating gender bias in ML-based clinical risk prediction models: A study on multiple use cases at different hospitals.

BACKGROUND: An inherent difference exists between male and female bodies, the historical under-representation of females in clinical trials widened this gap in existing healthcare data. The fairness of clinical decision-support tools is at risk when developed based on biased data. This paper aims to quantitatively assess the gender bias in risk prediction models. We aim to generalize our findings by performing this investigation on multiple use cases at different hospitals. METHODS: First, we conduct a thorough analysis of the source data to find gender-based disparities. Secondly, we assess the model performance on different gender groups at different hospitals and on different use cases. Performance evaluation is quantified using the area under the receiver-operating characteristic curve (AUROC). Lastly, we investigate the clinical implications of these biases by analyzing the underdiagnosis and overdiagnosis rate, and the decision curve analysis (DCA). We also investigate the influence of model calibration on mitigating gender-related disparities in decision-making processes. RESULTS: Our data analysis reveals notable variations in incidence rates, AUROC, and over-diagnosis rates across different genders, hospitals and clinical use cases. However, it is also observed the underdiagnosis rate is consistently higher in the female population. In general, the female population exhibits lower incidence rates and the models perform worse when applied to this group. Furthermore, the decision curve analysis demonstrates there is no statistically significant difference between the model's clinical utility across gender groups within the interested range of thresholds. CONCLUSION: The presence of gender bias within risk prediction models varies across different clinical use cases and healthcare institutions. Although inherent difference is observed between male and female populations at the data source level, this variance does not affect the parity of clinical utility. In conclusion, the evaluations conducted in this study highlight the significance of continuous monitoring of gender-based disparities in various perspectives for clinical risk prediction models.

SemiSynBio: A new era for neuromorphic computing.

Neuromorphic computing has the potential to achieve the requirements of the next-generation artificial intelligence (AI) systems, due to its advantages of adaptive learning and parallel computing. Meanwhile, biocomputing has seen ongoing development with the rise of synthetic biology, becoming the driving force for new generation semiconductor synthetic biology (SemiSynBio) technologies. DNA-based biomolecules could potentially perform the functions of Boolean operators as logic gates and be used to construct artificial neural networks (ANNs), providing the possibility of executing neuromorphic computing at the molecular level. Herein, we briefly outline the principles of neuromorphic computing, describe the advances in DNA computing with a focus on synthetic neuromorphic computing, and summarize the major challenges and prospects for synthetic neuromorphic computing. We believe that constructing such synthetic neuromorphic circuits will be an important step toward realizing neuromorphic computing, which would be of widespread use in biocomputing, DNA storage, information security, and national defense.

Pharmacokinetics and apparent Michaelis constant for metabolite conversion of sorafenib in healthy and hepatocellular carcinoma-bearing rats.

Aim: Investigation of the pharmacokinetics of sorafenib (SRF) in rats with hepatocellular carcinoma (HCC). Methods: A reproducible ultra-HPLC-MS method for simultaneous determination of serum SRF, N-hydroxymethyl sorafenib and N-demethylation sorafenib. Results: Both the maximum serum concentrations (2.5-times) and the area under the serum concentration-time curve from 0 h to infinity (4.5-times) of SRF were observed to be significantly higher, with a greater than 3.0-fold decrease in the clearance rate in the HCC-bearing rats compared with these values in healthy animals. Further study revealed approximately 3.8- and 3.2-times increases in the apparent Michaelis constant for N-hydroxymethyl sorafenib and N-demethylation sorafenib conversions in the HCC-bearing rats. Conclusion: The low efficiency for the SRF conversions was a key contributor to the increased serum concentrations of SRF.

[Box: see text].

Cut-off values of haemoglobin and clinical outcomes in incident peritoneal dialysis: the PDTAP study.

BACKGROUND: To explore the cut-off values of haemoglobin (Hb) on adverse clinical outcomes in incident peritoneal dialysis (PD) patients based on a national-level database. METHODS: The observational cohort study was from the Peritoneal Dialysis Telemedicine-assisted Platform (PDTAP) dataset. The primary outcomes were all-cause mortality, major adverse cardiovascular events (MACE) and modified MACE (MACE+). The secondary outcomes were the occurrences of hospitalization, first-episode peritonitis and permanent transfer to haemodialysis (HD). RESULTS: A total of 2591 PD patients were enrolled between June 2016 and April 2019 and followed up until December 2020. Baseline and time-averaged Hb <100 g/l were associated with all-cause mortality, MACE, MACE+ and hospitalizations. After multivariable adjustments, only time-averaged Hb <100 g/l significantly predicted a higher risk for all-cause mortality {hazard ratio [HR] 1.83 [95% confidence interval (CI) 1.19-281], P = .006}, MACE [HR 1.99 (95% CI 1.16-3.40), P = .012] and MACE+ [HR 1.77 (95% CI 1.15-2.73), P = .010] in the total cohort. No associations between Hb and hospitalizations, transfer to HD and first-episode peritonitis were observed. Among patients with Hb ≥100 g/l at baseline, younger age, female, use of iron supplementation, lower values of serum albumin and renal Kt/V independently predicted the incidence of Hb <100 g/l during the follow-up. CONCLUSION: This study provided real-world evidence on the cut-off value of Hb for predicting poorer outcomes through a nation-level prospective PD cohort.

Amino acids analysis reveals serum methionine contributes to diagnosis of the Kawasaki disease in mice and children.

BACKGROUND: Kawasaki disease (KD) patients often lack early and definitive diagnosis due to insufficient clinical criteria, whereas biomarkers might accelerate the diagnostic process and treatment. METHODS: The KD mouse models were established and thirteen amino acids were determined. A total of 551 serum samples were collected including KD patients (n = 134), HCs (n = 223) and KD patients after intravascular immunoglobulin therapy (IVIG, n = 194). A paired analysis of pre- and post-IVIG was employed in 10 KD patients. RESULTS: The pathological alterations of the aorta, myocardial interstitium and coronary artery vessel were observed in KD mice; the serum levels of methionine in KD mice (n = 40) were markedly altered and negatively correlated with the C-reactive protein levels. Consistent with the mouse model, serum methionine were significantly decreased in KD children, with the relative variation ratio of KD with HCs above 30% and AUROC value of 0.845. Serum methionine were correlated with Z-Score and significantly restored to the normal ranges after KD patient IVIG treatment. Another case-control study with 10 KD patients with IVIG sensitivity and 20 healthy controls validated serum methionine as a biomarker for KD patients with AUROC of 0.86. Elevation of serum DNMT1 activities, but no differences of DNMT3a and DNMT3b, were observed in KD patients when comparing with those in the HCs. CONCLUSIONS: Our study validated that serum methionine was a potential biomarker for KD, the alteration of which is associated with the activation of DNMT1 in KD patients.

A novel dynamic cardiorespiratory coupling quantification method reveals the effect of aging on the autonomic nervous system.

Traditional cardiopulmonary coupling (CPC) based on the Fourier transform shares an inherent trade-off between temporal and frequency resolutions with fixed window designs. Therefore, a cross-wavelet cardiorespiratory coupling (CRC) method was developed to highlight interwave cardiorespiratory dynamics and applied to evaluate the age effect on the autonomic regulation of cardiorespiratory function. The cross-wavelet CRC visualization successfully reflected dynamic alignments between R-wave interval signal (RR intervals) and respiration. Strong and continuous CRC was shown if there was perfect temporal coordination between consecutive R waves and respiration, while CRC becomes weaker and intermittent without such coordination. Using real data collected on electrocardiogram (ECG) and respiratory signals, the heart rate variability (HRV) and CRC were calculated. Subsequently, comparisons were conducted between young and elderly individuals. Young individuals had significantly higher partial time and frequency HRV indices than elderly individuals, indicating stronger control of parasympathetic regulation. The overall coupling strength of the CRC of young individuals was higher than that of elderly individuals, especially in high-frequency power, which was significantly lower in the elderly group than in the young group, achieving better results than the HRV indices in terms of statistical significance. Further analyses of the time-frequency dynamics of CRC indices revealed that the coupling strength was consistently higher in the high-frequency (HF) band (0.15-0.4 Hz) in young participants compared to elderly individuals. The dynamic CRC between respiration and HRV indices was accessible by integrating the cross-wavelet spectrum and coherence. Young participants had a significantly higher level of CRC in the HF band, indicating that aging reduces vagus nerve modulation.

S1PR1 regulates ovarian cancer cell senescence through the PDK1-LATS1/2-YAP pathway.

Cell senescence deters the activation of various oncogenes. Induction of senescence is, therefore, a potentially effective strategy to interfere with vital processes in tumor cells. Sphingosine-1-phosphate receptor 1 (S1PR1) has been implicated in various cancer types, including ovarian cancer. The mechanism by which S1PR1 regulates ovarian cancer cell senescence is currently elusive. In this study, we demonstrate that S1PR1 was highly expressed in human ovarian cancer tissues and cell lines. S1PR1 deletion inhibited the proliferation and migration of ovarian cancer cells. S1PR1 deletion promoted ovarian cancer cell senescence and sensitized ovarian cancer cells to cisplatin chemotherapy. Exposure of ovarian cancer cells to sphingosine-1-phosphate (S1P) increased the expression of 3-phosphatidylinositol-dependent protein kinase 1 (PDK1), decreased the expression of large tumor suppressor 1/2 (LATS1/2), and induced phosphorylation of Yes-associated protein (p-YAP). Opposite results were obtained in S1PR1 knockout cells following pharmacological inhibition. After silencing LATS1/2 in S1PR1-deficient ovarian cancer cells, senescence was suppressed and S1PR1 expression was increased concomitantly with YAP expression. Transcriptional regulation of S1PR1 by YAP was confirmed by chromatin immunoprecipitation. Accordingly, the S1PR1-PDK1-LATS1/2-YAP pathway regulates ovarian cancer cell senescence and does so through a YAP-mediated feedback loop. S1PR1 constitutes a druggable target for the induction of senescence in ovarian cancer cells. Pharmacological intervention in the S1PR1-PDK1-LATS1/2-YAP signaling axis may augment the efficacy of standard chemotherapy.

AIDMAN: An AI-based object detection system for malaria diagnosis from smartphone thin-blood-smear images.

Malaria is a significant public health concern, with ∼95% of cases occurring in Africa, but accurate and timely diagnosis is problematic in remote and low-income areas. Here, we developed an artificial intelligence-based object detection system for malaria diagnosis (AIDMAN). In this system, the YOLOv5 model is used to detect cells in a thin blood smear. An attentional aligner model (AAM) is then applied for cellular classification that consists of multi-scale features, a local context aligner, and multi-scale attention. Finally, a convolutional neural network classifier is applied for diagnosis using blood-smear images, reducing interference caused by false positive cells. The results demonstrate that AIDMAN handles interference well, with a diagnostic accuracy of 98.62% for cells and 97% for blood-smear images. The prospective clinical validation accuracy of 98.44% is comparable to that of microscopists. AIDMAN shows clinically acceptable detection of malaria parasites and could aid malaria diagnosis, especially in areas lacking experienced parasitologists and equipment.

THE ALARMIN EFFECT OF HMGB1/RIP3 ON TRANSFUSION-RELATED ACUTE LUNG INJURY VIA TLR4/NF-ΚB OR MAPK PATHWAY.

ABSTRACT: Nonantibody-mediated transfusion-related acute lung injury (TRALI) may account for up to 25% of TRALI cases. This indicates the need for further research to understand the pathophysiological mechanisms involved beyond antibody mediation fully. During this research, a TRALI rat model was developed using the trauma-blood loss-massive transfusion method. The severity of pulmonary edema was checked via measurement of lung histopathological changes and the amount of Evans blue dye fluid and bronchoalveolar lavage fluid protein leakage. In addition, potential mechanisms of pathophysiological pathways and inflammation cascades were investigated in TRALI rats in vivo . The findings indicated that TRALI increased inflammatory cytokines and triggered elevated levels of high-mobility group box 1 (HMGB1)/receptor-interacting protein kinase 3 (RIP3), apoptosis protein, and mRNAs in the TM (TRALI model) group as opposed to the normal control. Furthermore, TRALI activated the toll-like receptor 4/nuclear factor kappa B and mitogen-activated protein kinase signaling pathways, which partially regulated the inflammatory response in the TRALI rats. A significant increase was observed in the inflammatory mediators HMGB1 and RIP3 during the early stages of TRALI, suggesting that these mediators could be used as diagnostic markers for TRALI. In addition, HMGB1 and RIP3 promoted the inflammatory response by stimulating the toll-like receptor 44/nuclear factor kappa B and mitogen-activated protein kinase signaling pathways in the lung tissue of rats. Identifying efficient agents from inflammatory mediators such as alarmin can be an innovative scheme for diagnosing and preventing TRALI. These findings give HMGB1 and RIP3 a strong theoretical and experimental foundation for clinical use.

Long-term weathering difference in soil-like indicators of bauxite residue mediates the multifunctionality driven by microbial communities.

Long-term weathering enhances the stability of ecosystem services and alters the microbiome, however, its influences on the relationship between microbial diversity and multifunctionality are still poorly understood. Hereby, 156 samples (0-20 cm) from five artificially divided functional zones including central bauxite residue zone (BR), the zone near residential area (RA), the zone near dry farming area (DR), the zone near natural forest area (NF), and the zone near grassland and forest area (GF) were collected in a typical disposal area to determine the heterogeneity and development of biotic and abiotic properties of bauxite residue. Residues in BR and RA exhibited higher values of pH, EC, heavy metals, and exchangeable sodium percentage compared to those in NF and GF. Our results showed a positive correlation between multifunctionality and soil-like quality during long-term weathering. Microbial diversity and microbial network complexity responded positively to multifunctionality within the microbial community, which was parallel with ecosystem functioning. Long-term weathering promoted oligotrophs-dominated bacterial assemblages (mostly Acidobacteria and Chloroflexi) and suppressed copiotrophs (including Proteobacteria and Bacteroidota), while the response of fungal communities was lower. Rare taxa from bacterial oligotrophs were particularly important at the current stage for maintaining ecosystem services and ensuring microbial network complexity. Our results underscore the significance of microbial ecophysiological strategies in response to changes in multifunctionality during long-term weathering, and highlight the necessity of conserving and augmenting the abundance of rare taxa to ensure the stable provision of ecosystem functions in bauxite residue disposal areas.

Root architectures differentiate the composition of organic carbon in bauxite residue during natural vegetation.

Understanding plant root architectures induced changes in organic carbon accumulation and conversion is critical to predicting carbon cycling and screening appropriate plant species for ecological restoration on bauxite residue disposal areas. According to the ecological investigation of a weathered bauxite residue disposal area, three plants with different root architectures including Artemisia lavandulaefolia (A. lavandulaefolia), moss, and Zanthoxylum simulans (Z. simulans) were selected to investigate the rhizosphere effects on the composition and structure of organic carbon in bauxite residue. The physic-chemical properties, the contents and structure of different organic carbon fractions, and microbial communities of bauxite residue from rhizosphere and non-rhizosphere were analyzed. Plant growth decreased the saline-alkalinity, increased the contents of total organic carbon, particulate organic carbon and dissolved organic carbon, whilst enhancing the enzymatic activities of bauxite residue. Meanwhile, the rhizosphere effects had significant effects on the accumulation and stabilization of organic carbon in bauxite residue. A. lavandulaefolia had the strongest rhizosphere effects on the composition and structure of total organic carbon and dissolved organic carbon, whilst moss was more effective on the accumulation of particulate organic carbon in bauxite residue. Plant growth and root architecture changed the abundance of specific functional microorganisms and the complexity of microbial co-occurrence networks, thus elevating organic carbon levels in bauxite residue. During natural vegetation encroachment, rhizosphere exciting effects of the salt-tolerated plants could change the composition and structure of organic carbon fractions due to the comprehensive effectiveness of the improvement of physic-chemical properties and microbial communities. The findings improve our understanding of the responses of sequestration and stabilization of organic carbon pools to ecological restoration on bauxite residue disposal areas.

Irregular Scene Text Detection Based on a Graph Convolutional Network.

Detecting irregular or arbitrary shape text in natural scene images is a challenging task that has recently attracted considerable attention from research communities. However, limited by the CNN receptive field, these methods cannot directly capture relations between distant component regions by local convolutional operators. In this paper, we propose a novel method that can effectively and robustly detect irregular text in natural scene images. First, we employ a fully convolutional network architecture based on VGG16_BN to generate text components via the estimated character center points, which can ensure a high text component detection recall rate and fewer noncharacter text components. Second, text line grouping is treated as a problem of inferring the adjacency relations of text components with a graph convolution network (GCN). Finally, to evaluate our algorithm, we compare it with other existing algorithms by performing experiments on three public datasets: ICDAR2013, CTW-1500 and MSRA-TD500. The results show that the proposed method handles irregular scene text well and that it achieves promising results on these three public datasets.

Orthogonal threading-through-ß-sheet design of lung cancer EGFR extracellular domain-derived peptidic mimotopes binding to anti-EGFR antibody.

Human epidermal growth factor receptor (EGFR) has been established as a therapeutic target of lung cancer and other diverse tumors. The antibody drug Cetuximab has been developed to target the third subdomain III (TSDIII) of EGFR extracellular domain (ECD) by competitively inhibiting epidermal growth factor binding. In this study, we performed systematic investigation on the crystal complex structure of EGFR ECD domain with Cetuximab to create a residue importance profile for the TSDIII subdomain, based on which a number of U-shaped, double-stranded linear peptides were derived and cyclized to orthogonally thread through most hotspot residues and many responsible residues within the TSDIII ß-sheet plane; they represent mimotopes of the key antibody-recognition site of TSDIII subdomain. Computational analyses revealed that these linear peptides cannot spontaneously fold to the desired conformation in free state due to their intrinsic flexibility. Cell-free assays confirmed that the stapling can considerably improve the binding affinity of linear peptides to Cetuximab by up to 18-fold. The cOrt1 [3-18] cyclic peptide was measured to have the highest affinity in all designed linear and cyclic peptides.

Telemedicine and Clinical Outcomes in Peritoneal Dialysis: A Propensity-Matched Study.

INTRODUCTION: Telemedicine (TM) has shown to provide potential benefits on clinical outcomes in patients with chronic kidney disease but limited evidences published in the peritoneal dialysis (PD) population. This study aimed to explore the long-term effects of TM on the mortality and technique failure. METHODS: The Peritoneal Dialysis Telemedicine-assisted Platform Cohort Study (PDTAP Study) was conducted prospectively in 27 hospitals in China since 2016. Patient and practice data were collected through the doctor-end of the TM app (Manburs) for all participants. TM including self-monitoring records, on-line education materials, and real-time physician-patient contact was only performed for the patient-end users of the Manburs. The primary outcome was all-cause mortality. The secondary outcomes were cause-specific mortality and all-cause and cause-specific permanent transfer to hemodialysis. RESULTS: A total of 7,539 PD patients were enrolled between June 2016 and April 2019, with follow-up till December 2020. Patients were divided into two cohorts: TM group (39.1%) and non-TM group (60.9%). A propensity score was used to create 2,160 matched pairs in which the baseline covariates were well-balanced. There were significantly lower risks of all-cause mortality (HR 0.59 [0.51, 0.67], p < 0.001), CVD mortality (HR 0.59 [0.49, 0.70], p < 0.001), all-cause transfer to hemodialysis (0.57 [0.48, 0.67], p < 0.001), transfer to hemodialysis from PD-related infection (0.67 [0.51, 0.88], p = 0.003), severe fluid overload (0.40 [0.30, 0.55], p < 0.001), inadequate solute clearance (0.49 [0.26, 0.92], p = 0.026), and catheter-related noninfectious complications (0.41 [0.17, 0.97], p = 0.041) in the TM group compared with the non-TM group. CONCLUSION: This study indicated real-world associations between TM usage and reduction in patient survival and technique survival through a multicenter prospective cohort.

Self-Supervised Deep Multiview Spectral Clustering.

Multiview spectral clustering has received considerable attention in the past decades and still has great potential due to its unsupervised integration manner. It is well known that pairwise constraints boost the clustering process to a great extent. Nevertheless, the constraints are usually marked by human beings. To ameliorate the performance of multiview spectral clustering and alleviate the consumption of human resources, we propose self-supervised multiview spectral clustering with a small number of automatically retrieved pairwise constraints. First, the fused multiple autoencoders are used to extract the latent consistent feature of multiple views. Second, the pairwise constraints are achieved based on the commonality among multiple views. Then, the pairwise constraints are propagated through the neural network with historical memory. Finally, the propagated constraints are used to optimize the fused affinity matrix of spectral clustering. Our experiments on four benchmark datasets show the effectiveness of our proposed approach.

CpG-binding protein CFP1 promotes ovarian cancer cell proliferation by regulating BST2 transcription.

Epigenetic alterations have been functionally linked to ovarian cancer development and occurrence. The CXXC zinc finger protein 1 (CFP1) is an epigenetic regulator involved in DNA methylation and histone modification in mammalian cells. However, its role in ovarian cancer cells is unknown. Here, we show that CFP1 protein is highly expressed in human ovarian cancer tissues. Loss of CFP1 inhibited the growth of human ovarian cancer cells, promoted apoptosis, and increased senescence. CFP1 knockdown resulted in reduced levels of SETD1 (a CFP1 partner) and histone H3 trimethylation at the fourth lysine residue (H3K4me3). RNA-sequencing revealed that deletion of CFP1 resulted in mRNA reduction of bone marrow stromal cell antigen 2 (BST2). Bioinformatics analysis and chromatin immunoprecipitation showed that CFP1 binds to the promoter of BST2 and regulates its transcription directly. Overexpression of BST2 rescued the growth inhibitory effect of CFP1 loss. Furthermore, depletion of cullin-RING ubiquitin ligases 4 (CRL4) components ROC1 or CUL4A had significantly inhibited the expression of CFP1 and BST2 similar to MLN4924 treatment that blocked cullin neddylation and inactivated CRL4s. In conclusion, CFP1 promotes ovarian cancer cell proliferation and apoptosis by regulating the transcription of BST2, and the expression of CFP1 was affected by CRL4 ubiquitin ligase complex.

Attributed graph clustering with multi-task embedding learning.

Attributed graph clustering is challenging as it needs to effectively combine both graph structure and node feature information to accomplish node clustering. Recent studies mostly adopt graph neural networks to learn node embeddings, then apply traditional clustering methods to obtain clusters. However, their node embeddings are not specifically designed for clustering. Moreover, most of their loss functions only rely on either structure or feature information, making both kinds of information not fully retained in node embeddings. In this paper, we propose a multi-task embedding learning method (MTEL) for attributed graph clustering, which constructs two prediction tasks in terms of structure and feature based adjacency matrices respectively. To make the node embeddings helpful for the downstream clustering, in each task, we predict the minimum hop number between each pair of nodes in the adjacency matrix, so that the correlation degrees among nodes can be encoded into node embeddings. To improve the performance of the prediction task, we regularize the model parameters in these two tasks via ℓ2,1 norm, through which the model parameters can be jointly learned. Experiments on real attributed graphs show that MTEL is superior for attributed graph clustering over state-of-the-art methods.

Flexible pressure sensors via engineering microstructures for wearable human-machine interaction and health monitoring applications.

Flexible pressure sensors capable of transducing pressure stimuli into electrical signals have drawn extensive attention owing to their potential applications for human-machine interaction and healthcare monitoring. To meet these application demands, engineering microstructures in the pressure sensors are an efficient way to improve key sensing performances, such as sensitivity, linear sensing range, response time, hysteresis, and durability. In this review, we provide an overview of the recent advances in the fabrication and application of high-performance flexible pressure sensors via engineering microstructures. The implementation mechanisms and fabrication strategies of microstructures including micropatterned, porous, fiber-network, and multiple microstructures are systematically summarized. The applications of flexible pressure sensors with microstructures in the fields of wearable human-machine interaction, and ex vivo and in vivo healthcare monitoring are comprehensively discussed. Finally, the outlook and challenges in the future improvement of flexible pressure sensors toward practical applications are presented.

Graph Clustering With Graph Capsule Network.

Graph clustering, which aims to partition a set of graphs into groups with similar structures, is a fundamental task in data analysis. With the great advances made by deep learning, deep graph clustering methods have achieved success. However, these methods have two limitations: (1) they learn graph embeddings by a neural language model that fails to effectively express graph properties, and (2) they treat embedding learning and clustering as two isolated processes, so the learned embeddings are unsuitable for the subsequent clustering. To overcome these limitations, we propose a novel capsule-based graph clustering (CGC) algorithm to cluster graphs. First, we construct a graph clustering capsule network (GCCN) that introduces capsules to capture graph properties. Second, we design an iterative optimization strategy to alternately update the GCCN parameters and clustering assignment parameters. This strategy leads GCCN to learn cluster-oriented graph embeddings. Experimental results show that our algorithm achieves performance superior to that of existing graph clustering algorithms in terms of three standard evaluation metrics: ACC, NMI, and ARI. Moreover, we use visualization results to analyze the effectiveness of the capsules and demonstrate that GCCN can learn cluster-oriented embeddings.

Improvements on physical conditions of bauxite residue following application of organic materials.

Soil formation and ecological rehabilitation is the most promising strategy to eliminate environmental risks of bauxite residue disposal areas. Its poor physical structure is nevertheless a major limitation to plant growth. Organic materials were demonstrated as effective ameliorants to improve the physical conditions of bauxite residue. In this study, three different organic materials including straw (5% W/W), humic acid (5% W/W), and humic acid-acrylamide polymer (0.2% and 0.4%, W/W) were selected to evaluate their effects on physical conditions of bauxite residue pretreated by phosphogypsum following a 120-day incubation experiment. The proportion of 2-1 mm macro-aggregates, mean weight diameter (MWD) and geometric mean diameter (GWD) increased following organic materials addition, which indicated that organic materials could enhance aggregate stability. Compared with straw, and humic acid, humic acid-acrylamide polymer application had improved effects on the formation of water-stable aggregates in the residues. Furthermore, organic materials increased the total porosity, total pore volume and average pore diameter, and reduced the micropore content according to nitrogen gas adsorption (NA) and mercury intrusion porosimetry (MIP) analysis, whilst enhancing water retention of the residues based on water characteristic curves. Compared with traditional organic wastes, humic acid-acrylamide polymer could be regarded as a candidate according to the comprehensive consideration of the additive amount and the effects on physical conditions of bauxite residue. These findings could provide a novel application to both Ca-contained acid solid waste and high-molecular polymers on ecological rehabilitation at disposal areas.