Machine learning-driven prediction of phosphorus removal performance of metal-modified biochar and optimization of preparation processes considering water quality management objectives.

Based on water quality conditions and management, developing an optimized and targeted design approach for metal-modified biochar is achievable through machine learning. This study leveraged machine learning to analyze experimental data on phosphate adsorption by metal-modified biochar from literature published in Web of Science during 2014-2023. Using six machine learning models, phosphate adsorption capacity of biochar and residual phosphate concentration were predicted. Following hyperparameter optimization, gradient boosting model exhibited superior training performance (R2 > 0.96). Metal load quantity, solid-liquid ratio, and pH are key factors influencing adsorption performance. Optimal preparation parameters indicated that Mg-modified biochar achieved the highest adsorption capacity (387-396 mg/g), while La-modified biochar displayed the lowest residual phosphate concentration (0 mg/L). The results of verification experiments based on optimized process parameters closely aligned with model predictions. This research introduces a new machine learning-based approach for tailoring biochar preparation processes considering different water quality management objectives.

Green synthesis of CaxLa1-xMnO3 with modulation of mesoporous and vacancies for efficient low concentration phosphate adsorption.

Phosphate concentrations in eutrophic surface waters are usually low, and efficient removal of low concentration phosphate remains a challenge. In this study, Ca-doped LaMnO3 synthesized at doping ratios, designated as CaxLa1-xMnO3 (x = 0, 0.2, 0.4, 0.7), were compared. It was found that, the adsorption capacity of Ca0.4La0.6MnO3 material reached 63.01 mg/g at pH = 5, increased by 63.6% over the undoped LaMnO3 perovskite. For long-term adsorption, Ca0.4La0.6MnO3 could constantly adsorb phosphate to avoid phosphate accumulation (<0.05 mg/L). This proves that Ca0.4La0.6MnO3 has the ability to control dynamic water eutrophication. Characterization and density functional theory results confirmed that CaxLa1-xMnO3 can increase the content of mesopores and oxygen vacancies, providing additional active sites. This reduces the adsorption energy of the La site, promotes electron transfer, and increases its affinity. It provides a new method for removing low-concentration phosphates.

B-site metal modulation of phosphate adsorption properties and mechanism of LaBO3 (B = Fe, Al and Mn) perovskites.

La-based adsorbents are widely used for controlling phosphate concentration in water bodies. In order to explore the effect of different B-site metals regulating La-based perovskites on phosphate adsorption, three La-based perovskites (LaBO3, B = Fe, Al, and Mn) were prepared using the citric acid sol-gel method. Adsorption experiments showed that LaFeO3 exhibited the highest adsorption capacity for phosphate, which was 2.7 and 5 times higher than those of LaAlO3 and LaMnO3, respectively. The characterization results demonstrated that LaFeO3 has dispersed particles exhibiting larger pore size and more pores than LaAlO3 and LaMnO3. Spectroscopy analysis and density functional theory calculation results showed that different B-positions cause a change in the type of perovskite crystals. Among them, the differences between lattice oxygen consumption ratio, zeta potential and adsorption energy are the main reasons for the differences in adsorption capacity. In addition, the adsorption of phosphate by La-based perovskites were well fitted with Langmuir isotherm and pursues the pseudo-second-order kinetic models. The maximum adsorption capacities were 33.51, 12.31 and 6.61 mg/g for LaFeO3, LaAlO3 and LaMnO3, respectively. The adsorption mechanism was mainly based on inner-sphere complexation and electrostatic attraction. This study provides an explanation for the influence of different B sites on phosphate adsorption by perovskite.

Effects of a Modified Chitosan Compound Combined with Lung Lavage after Inhalation of Depleted Uranium Dust.

ABSTRACT: When exposed to depleted uranium (DU), the respiratory tract is the main route for DU to enter the body. At present, lung lavage is considered to be a method for removing DU from the lung. However, there is still room for improvement in the efficiency of lung lavage. In this work, a model of DU dust inhalation injury was established in beagle dogs so that chitosan-diethylenetriaminepentaacetic nanoparticles (CS-DTPA NP) could be synthesized. The purpose of this work was to evaluate the removal efficiency of CS-DTPA NP combined with lung lavage in dogs. Results showed that 7 d after DU exposure, the diethylenetriaminepentaacetic (DTPA) and CS-DTPA NP groups showed lower U content in kidney tissue compared with the normal saline (NS) group. In the left lung tissue (lavage fluid and recovery rate of lavage fluid), the U content in the CS-DTPA NP group was higher than in the NS and DTPA groups. In terms of blood levels, the CS-DPTA NP group increased over time at 1, 3 and 7 d of DU exposure without lavage; however, the U levels in the 3 and 7 d lavage groups were significantly lower than in the non-lavage groups. IL-1 in the lavage fluid of the CS-DPTA NP and CS NPs group were lower than in the NS group. In summary, after respiratory exposure to DU, early inhalation of CS-DPTA NP may block insoluble DU particles in the lung, and if combined with lung lavage, the clearance efficiency of DU from lung tissue improves.

Effect of early whole lung lavage at different time-points for promoting the removal of depleted uranium from the lung.

PURPOSE: This study compared the effect of whole lung lavage (WLL) at different time-points early after exposure of the respiratory system to insoluble radioactive particles. MATERIALS AND METHODS: Forty adult beagles were randomized into a control group and the 3-h, 8-h, 24-h, and 48-h lavage groups (n = 8). A canine model of acute lung injury was established by spraying a depleted uranium (DU) suspension using a superfine fiber bronchoscope, at a dose of 20 mg/kg. The lavage groups were subjected to WLL at 3 h, 8 h, 24 h, and 48 h post-DU exposure, while the control group received no treatment after exposure. Measurement of U in serum was performed using inductively coupled plasma mass spectrometry; measurements in the lavage fluid and left lung tissue were performed using inductively coupled plasma atomic emission spectrometry. The color of the lavage fluid was analyzed using colorimetry, and shadow changes in the lung were observed using chest computed tomography (CT). RESULTS: The lavage groups showed similarly increasing trends for serum U levels from DU exposure to 3 and 7 days after exposure; however, these values were significantly lower than those in the control group (p < .01). The U content in the lavage fluid was significantly higher in the 3-h group than in the 8-h, 24-h, and 48-h groups (p < .01), while that in the 8-h group was markedly higher than those in the 24-h and 48-h groups (p < .05). The average clearance rate of DU in the lungs varied in the range of 0.63‒7.06%. The U content in the left lung tissue of each lavage group was significantly lower than that in the control group (p < .01), while the content in the 8-h, 24-h, and 48-h groups was significantly higher than that in the 3-h group (p < .05). The colorimetric score of the lavage fluid in the 3-h group was significantly lower than those in the 8-h, 24-h, and 48-h groups (p < .05). Chest CT showed different degrees of consolidation and ground glass shadow changes in all groups. The score of the left lung shadow volume in the 3-h group was significantly lower than in the control, 8-h, 24-h, and 48-h groups (p < .01), while the score in the 8-h group was significantly higher than those in the 48-h and control groups (p < .05). CONCLUSIONS: The best effect of WLL after exposure of the respiratory system to insoluble radioactive particles was achieved at 3 h, followed by 8 h; there was no difference in the effectiveness of lung lavage at 24 h and 48 h.

"Keep it simple, scholar": an experimental analysis of few-parameter segmentation networks for retinal vessels in fundus imaging.

PURPOSE: With the recent development of deep learning technologies, various neural networks have been proposed for fundus retinal vessel segmentation. Among them, the U-Net is regarded as one of the most successful architectures. In this work, we start with simplification of the U-Net, and explore the performance of few-parameter networks on this task. METHODS: We firstly modify the model with popular functional blocks and additional resolution levels, then we switch to exploring the limits for compression of the network architecture. Experiments are designed to simplify the network structure, decrease the number of trainable parameters, and reduce the amount of training data. Performance evaluation is carried out on four public databases, namely DRIVE, STARE, HRF and CHASE_DB1. In addition, the generalization ability of the few-parameter networks are compared against the state-of-the-art segmentation network. RESULTS: We demonstrate that the additive variants do not significantly improve the segmentation performance. The performance of the models are not severely harmed unless they are harshly degenerated: one level, or one filter in the input convolutional layer, or trained with one image. We also demonstrate that few-parameter networks have strong generalization ability. CONCLUSION: It is counter-intuitive that the U-Net produces reasonably good segmentation predictions until reaching the mentioned limits. Our work has two main contributions. On the one hand, the importance of different elements of the U-Net is evaluated, and the minimal U-Net which is capable of the task is presented. On the other hand, our work demonstrates that retinal vessel segmentation can be tackled by surprisingly simple configurations of U-Net reaching almost state-of-the-art performance. We also show that the simple configurations have better generalization ability than state-of-the-art models with high model complexity. These observations seem to be in contradiction to the current trend of continued increase in model complexity and capacity for the task under consideration.

Learning with Known Operators reduces Maximum Training Error Bounds.

We describe an approach for incorporating prior knowledge into machine learning algorithms. We aim at applications in physics and signal processing in which we know that certain operations must be embedded into the algorithm. Any operation that allows computation of a gradient or sub-gradient towards its inputs is suited for our framework. We derive a maximal error bound for deep nets that demonstrates that inclusion of prior knowledge results in its reduction. Furthermore, we also show experimentally that known operators reduce the number of free parameters. We apply this approach to various tasks ranging from CT image reconstruction over vessel segmentation to the derivation of previously unknown imaging algorithms. As such the concept is widely applicable for many researchers in physics, imaging, and signal processing. We assume that our analysis will support further investigation of known operators in other fields of physics, imaging, and signal processing.

Genomic variability and molecular evolution of Asian isolates of sugarcane streak mosaic virus.

Sugarcane streak mosaic virus (SCSMV), an economically important causal agent of mosaic disease of sugarcane, is a member of the newly created genus Poacevirus in the family Potyviridae. In this study, we report the molecular characterization of three new SCSMV isolates from China (YN-YZ211 and HN-YZ49) and Myanmar (MYA-Formosa) and their genetic variation and phylogenetic relationship to SCSMV isolates from Asia and the type members of the family Potyviridae. The complete genome of each of the three isolates was determined to be 9781 nucleotides (nt) in size, excluding the 3' poly(A) tail. Phylogenetic analysis of the complete polyprotein amino acid (aa) sequences (3130 aa) revealed that all SCSMV isolates clustered into a phylogroup specific to the genus Poacevirus and formed two distinct clades designated as group I and group II. Isolates YN-YZ211, HN-YZ49 and MYA-Formosa clustered into group I, sharing 96.8-99.5 % and 98.9-99.6 % nt (at the complete genomic level) and aa (at the polyprotein level) identity, respectively, among themselves and 81.2-98.8 % and 94.0-99.6 % nt (at the complete genomic level) and aa (at the polyprotein level) identity, respectively, with the corresponding sequences of seven Asian SCSMV isolates. Population genetic analysis revealed greater between-group (0.190 ± 0.004) than within-group (group I = 0.025 ± 0.001 and group II = 0.071 ± 0.003) evolutionary divergence values, further supporting the results of the phylogenetic analysis. Further analysis indicated that natural selection might have contributed to the evolution of isolates belonging to the two identified SCSMV clades, with infrequent genetic exchanges occurring between them over time. These findings provide a comprehensive analysis of the population genetic structure and driving forces for the evolution of SCSMV with implications for global exchange of sugarcane germplasm.

A One-Step Real-Time RT-PCR Assay for the Detection and Quantitation of Sugarcane Streak Mosaic Virus.

Sugarcane mosaic disease is caused by the Sugarcane streak mosaic virus (SCSMV; genus Poacevirus, family Potyviridae) which is common in some Asian countries. Here, we established a protocol of a one-step real-time quantitative reverse transcription PCR (real-time qRT-PCR) using the TaqMan probe for the detection of SCSMV in sugarcane. Primers and probes were designed within the conserved region of the SCSMV coat protein (CP) gene sequences. Standard single-stranded RNA (ssRNA) generated by PCR-based gene transcripts of recombinant pGEM-CP plasmid in vitro and total RNA extracted from SCSMV-infected sugarcane were used as templates of qRT-PCR. We further performed a sensitivity assay to show that the detection limit of the assay was 100 copies of ssRNA and 2 pg of total RNA with good reproducibility. The values obtained were approximately 100-fold more sensitive than those of the conventional RT-PCR. A higher incidence (68.6%) of SCSMV infection was detected by qRT-PCR than that (48.6%) with conventional RT-PCR in samples showing mosaic symptoms. SCSMV-free samples were verified by infection with Sugarcane mosaic virus (SCMV) or Sorghum mosaic virus (SrMV) or a combination of both. The developed qRT-PCR assay may become an alternative molecular tool for an economical, rapid, and efficient detection and quantification of SCSMV.