Mineralogy and Geochemistry of Rare Earth Elements in Carboniferous-Permian Coals at the Eastern Margin of the Ordos Basin.

This article used Carboniferous-Permian coals from the Jungar, Hedong, and Weibei Coalfields in the east of the Ordos Basin as research samples. Characteristics of coal quality, petrology, mineralogy, and geochemistry were analyzed by proximate analysis, inductively coupled plasma mass spectrometry, X-ray fluorescence spectroscopy, X-ray diffraction analysis, scanning electron microscopy-energy spectrum analysis, and incident light microscope. The enrichment regulations, distribution patterns, and occurrences of REY (rare earth element and yttrium) in coal under different geological conditions were compared. Geological significance and the influence of REY were then discussed. The average REY of Permian coal in the eastern margin of the basin is 127.9 µg/g, CC = 1.87, and the average REY of Carboniferous coal is 117.49 µg/g, CC = 1.72, which are within the normal enrichment range. The inorganic affinity of REYs in the study area is strong and mainly occurs in clay minerals and detrital phosphates and correlates well with LREY. The Permian coal sedimentary environment is more oxidized than the Taiyuan formation, and the Carboniferous coal sedimentary environment is noticeably more affected by marine water. With an increasing degree of coalification, the concentration of rare earth elements (REE) in high-rank coal vitrinite is lower than that in inertinite. In contrast, the concentration of REEs in low-rank coal is the opposite. This is because the oxygen-containing functional groups that can combine with REEs in vitrinite reduce significantly, resulting in the loss of trace elements into other forms. The provenance of the northern and central regions of the study area is mainly sedimentary rocks, granite, alkaline basalt, and continental tholeiite, while the southern region is mainly granite and sedimentary rocks.

The developments of C-N bond formation via electrochemical Ritter-type reactions.

With the development of organic electrochemical synthesis, a series of notable achievements have been made in electrochemical Ritter amination reactions, which have enriched the methods available for constructing C-N bonds. In this review, electrochemical Ritter amination reactions are introduced based on the classification of reaction substrates, including olefins, aromatics, alkylbenzenes, and the less reported carboxylic acids, ketones, sulfides, and alkanes. The application of electrochemical technology to Ritter reactions has improved the harsh conditions of the traditional reactions, and extended the substrate scope and the structural diversity of the products. The application value of Ritter reactions in organic synthesis has also been further expanded.

Biologically interpretable multi-task deep learning pipeline predicts molecular alterations, grade, and prognosis in glioma patients.

Deep learning models have been developed for various predictions in glioma; yet, they were constrained by manual segmentation, task-specific design, or a lack of biological interpretation. Herein, we aimed to develop an end-to-end multi-task deep learning (MDL) pipeline that can simultaneously predict molecular alterations and histological grade (auxiliary tasks), as well as prognosis (primary task) in gliomas. Further, we aimed to provide the biological mechanisms underlying the model's predictions. We collected multiscale data including baseline MRI images from 2776 glioma patients across two private (FAHZU and HPPH, n = 1931) and three public datasets (TCGA, n = 213; UCSF, n = 410; and EGD, n = 222). We trained and internally validated the MDL model using our private datasets, and externally validated it using the three public datasets. We used the model-predicted deep prognosis score (DPS) to stratify patients into low-DPS and high-DPS subtypes. Additionally, a radio-multiomics analysis was conducted to elucidate the biological basis of the DPS. In the external validation cohorts, the MDL model achieved average areas under the curve of 0.892-0.903, 0.710-0.894, and 0.850-0.879 for predicting IDH mutation status, 1p/19q co-deletion status, and tumor grade, respectively. Moreover, the MDL model yielded a C-index of 0.723 in the TCGA and 0.671 in the UCSF for the prediction of overall survival. The DPS exhibits significant correlations with activated oncogenic pathways, immune infiltration patterns, specific protein expression, DNA methylation, tumor mutation burden, and tumor-stroma ratio. Accordingly, our work presents an accurate and biologically meaningful tool for predicting molecular subtypes, tumor grade, and survival outcomes in gliomas, which provides personalized clinical decision-making in a global and non-invasive manner.

Prevalence and Prognostic Significance of Malnutrition in Patients with Type B Aortic Dissection Undergoing Endovascular Repair.

Background: Malnutrition is a poor prognostic factor in a wide range of diseases. Nevertheless, there is a lack of data investigating the association between malnutrition and outcomes of patients with type B aortic dissection (TBAD) undergoing thoracic endovascular aortic repair (TEVAR). Therefore, the aim of the present study was to report the prevalence and clinical impact of malnutrition assessed by the controlling nutritional status (CONUT) score in TBAD patients undergoing TEVAR. Methods: The retrospective study indicated that a total of 881 patients diagnosed with TBAD and treated with TEVAR from January 2010 to December 2017 were categorized into subgroups based on their CONUT score (low ≤ 5 vs. high > 5). To assess the correlation between malnutrition and early and follow-up outcomes of TBAD patients, logistic and Cox regression analysis were utilized, incorporating inverse probability weighting. Results: Malnutrition was present in 20.3% of patients according to the CONUT score. Multivariate logistic regression analysis revealed that pre-operative CONUT score modeled as a continuous variable was an independent risk factor for prolonged intensive care unit stay (odds ratio [OR], 1.09; 95% confidence interval [CI], 1.02-1.17; p = 0.015), 30-day death (OR, 1.43; 95% CI, 1.19-1.72; p < 0.001), delirium (OR, 1.11; 95% CI, 1.01-1.23; p = 0.035) and acute kidney injury (OR, 1.09; 95% CI, 1.01-1.16; p = 0.027). During a median follow-up of 70.8 (46.1-90.8) months, 102 (11.8%) patients died (high CONUT group: 21.8% vs. low CONUT group: 9.0%; p < 0.001). Multivariable Cox proportional-hazards models showed that malnutrition was an independent predictor for follow-up mortality (hazard ratio, 1.68; 95% CI, 1.11-2.53; p = 0.014). Results remained consistent across various sensitivity analyses. Conclusions: Malnutrition assessed by the CONUT score could profoundly affect the early and follow-up prognosis in patients undergoing TEVAR. Routine pre-intervention nutritional evaluation might provide valuable prognostic information.

Genome-Wide Identification and Expression Analysis of Growth-Regulating Factor Family in Sweet Potato and Its Two Relatives.

Growth-regulating factor (GRF) is a multi-gene family that plays an important role in plant growth and development and is widely present in plants. Currently, GRF gene members have been reported in many plants, but the GRF gene family has not been found in sweet potato. In this study, ten GRF genes were identified in sweet potato (Ipomoea batatas), twelve and twelve were identified in its two diploid relatives (Ipomoea trifida) and (Ipomoea triloba), which were unevenly distributed on nine different chromosomes. Subcellular localization analysis showed that GRF genes of sweet potato, I. trifida, and I. triloba were all located in the nucleus. The expression analysis showed that the expression of IbGRFs was diverse in different sweet potato parts, and most of the genes were upregulated and even had the highest expression in the vigorous growth buds. These findings provide molecular characterization of sweet potato and its two diploid relatives, the GRF families, further supporting functional characterization.

Distribution and modes of occurrence of Nb in subbituminous coal: A case study from the Jungar Coalfield, Ordos Basin, China.

Niobium has been considered to be enriched in high-Al-Ga in north China coal and coal-hosted Nb(Ta)-Zr(Hf)-REY-Ga polymetallic deposits in the southwestern region of China. However, modes of occurrence and influencing factors of Nb in Al-Ga-rich coal in North China are rarely reported. This study investigated the distribution characteristics of Nb in the No.6 high-Al-Ga coal of the Jungar Coalfield in North China. The in-situ Nb concentration of selected minerals, including kaolinite, Ti-oxides, boehmite, and zircon, is further quantitatively characterized based on multiscale in-situ elemental analyses, including SEM-EDS and LA-ICP-MS spot and mapping analyses. The results showed that Nb is rich in the tonstein and mudstone partings among bulk samples and is highly concentrated in Ti-oxides, followed by zircon among the minerals. A certain amount of Nb is associated with kaolinite and boehmite with different modes of occurrence: vermicular kaolinite (65.94 ppm) > clastic kaolinite (25.43 ppm) > altered K-bearing kaolinite (18.11 ppm) > cryptocrystalline kaolinite (17.03 ppm) > clastic boehmite (9.08 ppm) > cryptocrystalline boehmite (7.97 ppm). The Nb-Nb/Ta and Zr/Hf-Nb/Ta ratios suggest that the primary source of Nb in the No.6 coal is the altered felsic volcanic ash and bauxite deposit. The high-Fe concentration in Ti-bearing minerals indicates that anatase may originate from the alteration of ilmenite with a process of Fe depletion and Nb enrichment. The enrichment of Nb in coal is attributed to the substitution of Nb for Ti in isomorphism in all Ti-oxides, high-Ti, and Ti-bearing minerals.

A novel approach with tofacitinib for the management of keratoderma blennorrhagicum in reactive arthritis: a case report.

Reactive arthritis(ReA), a form of arthritis occurring post-infection, manifests with antecedent infection symptoms, arthritis, and extra-articular manifestations, categorizing it as spondyloarthritis. "Keratoderma blennorrhagicum" (characterized by pustular hyperkeratosis on palms and soles, resembling pustular psoriasis) represents the most typical skin manifestation of ReA, occurring in acute or chronic phases. Severe lesions necessitate systemic disease modifying anti-rheumatic drugs (DMARDs) or biologic therapies. This article reports a case of ReA with sacroiliitis and widespread pustular eruptions following a urinary tract infection. Treatment with sulfasalazine and thalidomide significantly improved sacroiliitis, but the skin rash remained persistent and recurring. Subsequent use of adalimumab and secukinumab resulted in worsening skin rash, prompting a switch to tofacitinib, leading to a remarkable improvement in pustular eruptions after 20 days of treatment. This case demonstrates successful application of tofacitinib in treating severe keratoderma blennorrhagicum refractory to conventional DMARDs and biologics, offering insights into JAK inhibition for challenging rheumatic diseases with skin involvement.

The mechanosensitive ion channel Piezo1 contributes to podocyte cytoskeleton remodeling and development of proteinuria in lupus nephritis.

Piezo1 functions as a special transducer of mechanostress into electrochemical signals and is implicated in the pathogenesis of various diseases across different disciplines. However, whether Piezo1 contributes to the pathogenesis of lupus nephritis (LN) remains elusive. To study this, we applied an agonist and antagonist of Piezo1 to treat lupus-prone MRL/lpr mice. Additionally, a podocyte-specific Piezo1 knockout mouse model was also generated to substantiate the role of Piezo1 in podocyte injury induced by pristane, a murine model of LN. A marked upregulation of Piezo1 was found in podocytes in both human and murine LN. The Piezo1 antagonist, GsMTx4, significantly alleviated glomerulonephritis and tubulointerstitial damage, improved kidney function, decreased proteinuria, and mitigated podocyte foot process effacement in MRL/lpr mice. Moreover, podocyte-specific Piezo1 deletion showed protective effects on the progression of proteinuria and podocyte foot process effacement in the murine LN model. Mechanistically, Piezo1 expression was upregulated by inflammatory cytokines (IL-6, TNF-α and IFN-γ), soluble urokinase Plasminogen Activator Receptor and its own activation. Activation of Piezo1 elicited calcium influx, which subsequently enhanced Rac1 activity and increased active paxillin, thereby promoting cytoskeleton remodeling and decreasing podocyte motility. Thus, our work demonstrated that Piezo1 contributed to podocyte injury and proteinuria progression in LN. Hence, targeted therapy aimed at decreasing or inhibiting Piezo1 could represent a novel strategy to treat LN.

Adenosine triggers early astrocyte reactivity that provokes microglial responses and drives the pathogenesis of sepsis-associated encephalopathy in mice.

Molecular pathways mediating systemic inflammation entering the brain parenchyma to induce sepsis-associated encephalopathy (SAE) remain elusive. Here, we report that in mice during the first 6 hours of peripheral lipopolysaccharide (LPS)-evoked systemic inflammation (6 hpi), the plasma level of adenosine quickly increased and enhanced the tone of central extracellular adenosine which then provoked neuroinflammation by triggering early astrocyte reactivity. Specific ablation of astrocytic Gi protein-coupled A1 adenosine receptors (A1ARs) prevented this early reactivity and reduced the levels of inflammatory factors (e.g., CCL2, CCL5, and CXCL1) in astrocytes, thereby alleviating microglial reaction, ameliorating blood-brain barrier disruption, peripheral immune cell infiltration, neuronal dysfunction, and depression-like behaviour in the mice. Chemogenetic stimulation of Gi signaling in A1AR-deficent astrocytes at 2 and 4 hpi of LPS injection could restore neuroinflammation and depression-like behaviour, highlighting astrocytes rather than microglia as early drivers of neuroinflammation. Our results identify early astrocyte reactivity towards peripheral and central levels of adenosine as an important pathway driving SAE and highlight the potential of targeting A1ARs for therapeutic intervention.

Unsupervised EEG-Based Seizure Anomaly Detection with Denoising Diffusion Probabilistic Models.

While many seizure detection methods have demonstrated great accuracy, their training necessitates a substantial volume of labeled data. To address this issue, we propose a novel method for unsupervised seizure anomaly detection called SAnoDDPM, which uses denoising diffusion probabilistic models (DDPM). We designed a novel pipeline that uses a variable lower bound on Markov chains to identify potential values that are unlikely to occur in anomalous data. The model is first trained on normal data, then anomalous data is input to the trained model. The model resamples the anomalous data and converts it to normal data. Finally, the presence of seizures can be determined by comparing the before and after data. Moreover, the input 2D spectrograms are encoded into vector-quantized representations, which enables powerful and efficient DDPM while maintaining its quality. Experimental comparisons on the publicly available datasets, CHB-MIT and TUH, show that our method delivers better results, significantly reduces inference time, and is suitable for deployment in a clinical environments. As far as we are aware, this is the first DDPM-based method for seizure anomaly detection. This novel approach significantly contributes to the progression of seizure detection algorithms, thereby augmenting their practicality in clinical settings.

Picosecond laser capture microdissection based on edge catapulting combined with dielectrophoretic force.

The spatial omics information analysis of heterogeneous cells or cell populations is of great importance for biomedical research. Herein, we proposed a picosecond laser capture microdissection boosted by edge catapulting combined with dielectrophoretic force (ps-LMED) that enables fast and non-invasive acquisition of uncontaminated cells and cell populations for downstream molecular assays. The target cells were positioned under a microscope and separated by a focused picosecond pulsed laser. The system employed the plasma expansion force during cutting to lift the target and captured it under dielectrophoretic force from the charged collection cap eventually. The principle of our system has been validated by both theoretical analysis and practical experiments. The results indicated that our system can collect samples ranging from a single cell with a diameter of a few microns to large tissues with a volume of 532,500 µm3 at the moment finishing the cutting, without further operations. The cutting experiments of living cells and ribonucleic acid (RNA) and protein omics analysis results of collected targets demonstrated the advantage of non-destructiveness to the samples and feasibility in omics applications.

Immunotherapy for recurrent or metastatic nasopharyngeal carcinoma.

Immunotherapy, particularly immune checkpoint inhibitors (ICIs), such as anti-programmed death 1/programmed death-ligand 1 (PD-1/PD-L1) therapy, has emerged as a pivotal treatment modality for solid tumors, including recurrent or metastatic nasopharyngeal carcinoma (R/M-NPC). Despite the advancements in the utilization of ICIs, there is still room for further improving patient outcomes. Another promising approach to immunotherapy for R/M-NPC involves adoptive cell therapy (ACT), which aims to stimulate systemic anti-tumor immunity. However, individual agent therapies targeting dendritic cells (DCs) appear to still be in the clinical trial phase. This current review underscores the potential of immunotherapy as a valuable adjunct to the treatment paradigm for R/M-NPC patients. Further research is warranted to enhance the efficacy of immunotherapy through the implementation of strategies such as combination therapies and overcoming immune suppression. Additionally, the development of a biomarker-based scoring system is essential for identifying suitable candidates for precision immunotherapy.

Performance evaluation of the effect of humic acid on Anammox granular sludge: Apparent morphology, nitrogen removal and microbial community.

Humic acid (HA) is a typical refractory organic matter, so it is of great significance to investigate its effect on the performance of Anammox granular sludge. When the dosage of HA ≤ 50 mg/L, HA promotes the total nitrogen removal rate (NRR) to 1.45 kg/(m3·day). When HA was between 50 and 100 mg/L, the NRR of Anammox was stable. At this time, the adsorption of HA causes the sludge to gradually turn from red to brown, but the activities of heme and enzymes showed that its capacity was not affected. When HA levels reached 250 mg/L, the NRR dropped to 0.11 kg/(m3·day). Moderate HA levels promoted the release of extracellular polymeric substance (EPS), but excessive HA levels lead to a decrease in EPS concentrations. HA inhibited Anammox activity, which indirectly hindered the transmission of substrate and accumulated substrate toxicity. Although HA promoted the increase of heterotrophic microbial abundance in Anammox system, the microbial diversity decreased gradually. With the increase of HA concentration, the abundance of Candidatus_Brocadia, the main functional microorganism of Anammox system, decreased gradually, while the abundance of Candidatus_Kuenenia increased gradually.

Nonalcoholic Fatty Liver Disease: Changes in Gut Microbiota and Blood Lipids.

Background and Aims: The global prevalence of nonalcoholic fatty liver disease (NAFLD) is 25%. This study aimed to explore differences in the gut microbial community and blood lipids between normal livers and those affected by NAFLD using 16S ribosomal deoxyribonucleic acid sequencing. Methods: Gut microbiome profiles of 40 NAFLD and 20 non-NAFLD controls were analyzed. Information about four blood lipids and 13 other clinical features was collected. Patients were divided into three groups by ultrasound and FibroScan, those with a normal liver, mild FL (FL1), and moderate-to-severe FL (FL2). FL1 and FL2 patients were divided into two groups, those with either hyperlipidemia or non-hyperlipidemia based on their blood lipids. Potential keystone species within the groups were identified using univariate analysis and a specificity-occupancy plot. Significant difference in biochemical parameters ion NAFLD patients and healthy individuals were identified by detrended correspondence analysis and canonical correspondence analysis. Results: Decreased gut bacterial diversity was found in patients with NAFLD. Firmicutes/Bacteroidetes decreased as NAFLD progressed. Faecalibacterium and Ruminococcus 2 were the most representative fatty-related bacteria. Glutamate pyruvic transaminase, aspartate aminotransferase, and white blood cell count were selected as the most significant biochemical indexes. Calculation of areas under the curve identified two microbiomes combined with the three biochemical indexes that identified normal liver and FL2 very well but performed poorly in diagnosing FL1. Conclusions: Faecalibacterium and Ruminococcus 2, combined with glutamate pyruvic transaminase, aspartate aminotransferase, and white blood cell count distinguished NAFLD. We speculate that regulating the health of gut microbiota may release NAFLD, in addition to providing new targets for clinicians to treat NAFLD.

Colchicine Blocks Abdominal Aortic Aneurysm Development by Maintaining Vascular Smooth Muscle Cell Homeostasis.

Development of non-surgical treatment of human abdominal aortic aneurysm (AAA) has clinical significance. Colchicine emerges as an effective therapeutic regimen in cardiovascular diseases. Yet, whether colchicine slows AAA growth remain controversy. Here, we demonstrated that daily intragastric administration of low-dose colchicine blocked AAA formation, prevented vascular smooth muscle cell (SMC) phenotype switching and apoptosis, and vascular inflammation in both peri-aortic CaPO4 injury and subcutaneous angiotensin-II infusion induced experimental AAA mice models. Mechanistically, colchicine increased global mRNA stability by inhibiting the METTL14/YTHDC1-mediated m6A modification, resulting in increased sclerostin (SOST) expression and consequent inactivation of the WNT/ß-catenin signaling pathway in vascular SMCs from mouse AAA lesions and in cultured human aortic SMCs. Moreover, human and mouse AAA lesions all showed increased m6A methylation, decreased SOST expression, and skewed synthetic SMC de-differentiation phenotype, compared to those without AAA. This study uncovers a novel mechanism of colchicine in slowing AAA development by using the METTL14/SOST/WNT/ß-catenin axis to control vascular SMC homeostasis in mouse aortic vessels and in human aortic SMCs. Therefore, use of colchicine may benefit AAA patients in clinical practice.

Protocol for embedded 3D printing of heart tissues using thiol-norbornene collagen.

Here, we present a protocol for 3D printing heart tissues using thiol-norbornene photoclick collagen (NorCol). We describe steps for synthesizing NorCol, preparing bioink and the support bath, and cell-laden printing. We then detail procedures for the loading of C2C12 cells into NorCol, ensuring structural integrity and cell viability after printing. This protocol is adaptable to various cell lines and allows for the printing of diverse complex structures, which can be used in drug screening and disease modeling.

[Prognostic risk factors of catheter-related bloodstream infection in patients with maintenance hemodialysis].

OBJECTIVE: To analyze the pathogen distribution and prognostic risk factors of catheter-related bloodstream infection (CRBSI) in patients with maintenance hemodialysis (MHD) during non-hospitalization. METHODS: A retrospective comparative study was conducted. Thirty-four patients of MHD with semi-permanent catheter admitted to the department of nephrology of Gansu Provincial Hospital from January 2020 to May 2023 due to CRBSI during non-hospitalization were enrolled. The distribution characteristics of pathogens causing CRBSI in MHD patients during non-hospital period were analyzed. All patients were actively given anti-infection treatment after admission. The general data, laboratory indicators and prognosis during hospitalization were collected through the electronic medical record system. Patients were divided into poor prognosis group (14 cases) and good prognosis group (20 cases) according to the treatment results during hospitalization. Univariate and binary Logistic regression were used to analyze the risk factors affecting the prognosis of patients, and receiver operator characteristic curve (ROC curve) was drawn to evaluate its predictive value for prognosis. RESULTS: A total of 28 pathogenic bacteria were isolated from 34 patients, of which 25 were Gram-positive, Staphylococcus was the most common pathogen, accounting for 82.15% of the total, and 16 strains of Staphylococcus aureus (57.15%), including 6 methicillin-resistant Staphylococcus aureus (MRSA, 21.43%). There were 7 strains of Staphylococcus epidermidis (25.00%), including 3 strains of methicillin-resistant Staphylococcus epidermidis (MRSE, 10.71%). There were 3 strains of Gram-negative bacteria, 1 strain each of Pseudomonas aeruginosa, Escherichia coli and Acinetobacter baumannii. Univariate analysis showed that the fever duration of MHD patients with CRBSI in the poor prognosis group was significantly longer than that in the good prognosis group [days: 8.50 (3.75, 45.00) vs. 2.50 (1.00, 4.75), P < 0.01], serum erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) and random blood glucose (GLU) were significantly higher than those in the good prognosis group [ESR (mm/1 h): 82.36±24.98 vs. 56.95±35.65, CRP (mg/L): 123.45±74.10 vs. 67.35±55.22, GLU (mmol/L): 8.74±3.66 vs. 6.42±1.95, all P < 0.05]. Binary Logistic regression analysis showed that serum CRP was an independent risk factor for poor prognosis in MHD patients with CRBSI [odds ratio (OR) = 1.020, 95% confidence interval (95%CI) was 1.002-1.038, P = 0.025]. ROC curve analysis showed that the area under the curve (AUC) of serum CRP in predicting poor prognosis of MHD patients with CRBSI was 0.711; when the optimal cut-off value was 104.65 mg/L, the sensitivity was 64.3% and the specificity was 85.0%, indicating that it has good predictive value. CONCLUSIONS: Gram-positive bacteria are the main pathogens of CRBSI in MHD patients during non-hospital period. The poor prognosis is mainly related to the high level of serum CRP. Serum CRP level can effectively screen the high-risk group of MHD patients with CRBSI with poor prognosis.

Peroxiredoxin 3 has a crucial role in the macrophage polarization by regulating mitochondrial homeostasis.

Acute lung injury (ALI) is one of the life-threatening complications of sepsis, and macrophage polarization plays a crucial role in the sepsis-associated ALI. However, the regulatory mechanisms of macrophage polarization in ALI and in the development of inflammation are largely unknown. In this study, we demonstrated that macrophage polarization occurs in sepsis-associated ALI and is accompanied by mitochondrial dysfunction and inflammation, and a decrease of PRDX3 promotes the initiation of macrophage polarization and mitochondrial dysfunction. Mechanistically, PRDX3 overexpression promotes M1 macrophages to differentiate into M2 macrophages, and enhances mitochondrial functional recovery after injury by reducing the level of glycolysis and increasing TCA cycle activity. In conclusion, we identified PRDX3 as a critical hub integrating oxidative stress, inflammation, and metabolic reprogramming in macrophage polarization. The findings illustrate an adaptive mechanism underlying the link between macrophage polarization and sepsis-associated ALI.

Velocity-aware spatial-temporal attention LSTM model for inverse dynamic model learning of manipulators.

Introduction: An accurate inverse dynamics model of manipulators can be effectively learned using neural networks. However, further research is required to investigate the impact of spatiotemporal variations in manipulator motion sequences on network learning. In this work, the Velocity Aware Spatial-Temporal Attention Residual LSTM neural network (VA-STA-ResLSTM) is proposed to learn a more accurate inverse dynamics model, which uses a velocity-aware spatial-temporal attention mechanism to extract dynamic spatiotemporal features selectively from the motion sequence of the serial manipulator. Methods: The multi-layer perception (MLP) attention mechanism is adopted to capture the correlation between joint position and velocity in the motion sequence, and the state correlation between hidden units in the LSTM network to reduce the weight of invalid features. A velocity-aware state fusion approach of LSTM network hidden units' states is proposed, which utilizes variation in joint velocity to adapt to the temporal characteristics of the manipulator dynamic motion, improving the generalization and accuracy of the neural network. Results: Comparative experiments have been conducted on two open datasets and a self-built dataset. Specifically, the proposed method achieved an average accuracy improvement of 61.88% and 43.93% on the two different open datasets and 71.13% on the self-built dataset compared to the LSTM network. These results demonstrate a significant advancement in accuracy for the proposed method. Discussion: Compared with the state-of-the-art inverse dynamics model learning methods of manipulators, the modeling accuracy of the proposed method in this paper is higher by an average of 10%. Finally, by visualizing attention weights to explain the training procedure, it was found that dynamic modeling only relies on partial features, which is meaningful for future optimization of inverse dynamic model learning methods.