This study investigates machine-learning-based prediction of bio-oil yield and quality indicators during the hydrothermal liquefaction (HTL) of lignocellulosic biomass. A curated literature-derived database was used to develop multilayer perceptron, support vector regression, decision tree, random forest (RF), and extreme gradient boosting (XGB) models for predicting bio-oil yield, elemental ratios, carbon retention, oxygen removal efficiency, and higher heating value. Predictor selection combined process knowledge with correlation analysis, while model hyperparameters were optimized using particle swarm optimization. XGB and RF generally provided the strongest performance; however, the models developed for most secondary outputs exhibited limited generalization. Consequently, detailed post-prediction analysis was restricted to bio-oil yield, for which the XGB model showed the most reliable predictive performance. Inclusion of the effective water fill ratio (Vw/Vr) substantially improved yield prediction by representing variations in solvent availability, reactor filling, and associated physicochemical interactions. SHapley Additive exPlanations ranked Vw/Vr as the most influential predictor, whereas Spearman analysis identified reaction time as the strongest monotonic factor (ρ ≈ −0.43), followed by Vw/Vr (ρ ≈ +0.34). Segmented regression and localized partial-dependence analyses reconciled these rankings by showing that the influence of reaction time was concentrated mainly below approximately 80 min, while Vw/Vr produced stronger nonlinear and regime-dependent changes, particularly below a model-derived breakpoint near 13.9. Joint-bootstrap Monte Carlo and residual-conformal analyses were further used to quantify predictive uncertainty and establish a model-credibility envelope based on interval width and applicability-domain support. Overall, the proposed framework provides an interpretable and uncertainty-aware approach for modeling lignocellulosic HTL and prioritizing conditions for subsequent process optimization.
Accurate quantification of retinal ganglion cells (RGCs) is critical for assessing neurodegeneration and evaluating therapeutic interventions in experimental glaucoma. Manual RGCs quantification is time-consuming, subjective, and inconsistent, underscoring the need for automated and scalable approaches that can adapt to varying cell densities and imaging conditions. We present XGB-UNetFuse (Fusion of Extreme Gradient Boosting (XGB) classifier with U-Net and Light-U-Net models) for automated RGCs quantification from microscopy images. The workflow begins with dataset preparation, including conversion of standard microscopic images into differential interference contrast (DIC)-like representations to enhance cell boundary visibility and contrast during preprocessing. Feature vectors are extracted and combined with manual ground truth to perform binary classification of whole images into low or moderate/high-density categories using an XGBoost model. Based on this classification, the system adaptively selects between two independently trained detection networks: Light-U-Net for low-density images and U-Net for moderate/high-density images. The outputs undergo tile reconstruction, Gaussian smoothing, intensity thresholding, and local-maxima detection to generate precise cell counts. Experimental validation across training, test, and an independent evaluation datasets demonstrated that XGB-UNetFuse achieved superior accuracy and generalization, yielding the lowest mean counting error and highest agreement with manual counts. Bland–Altman analysis confirmed a mean bias within ±10% across all retinal regions. The proposed density-aware adaptive framework provides a robust, scalable, and reproducible solution for high-throughput RGCs quantification, advancing automated retinal microscopic analysis and translational ophthalmic research.
This study reveals that the time-dependent hardening of dough is mainly driven by the cleavage and reformation of disulfide bonds. Tensile tests showed that dough treated with sodium metabisulfite (SMBS) had higher initial extensibility but hardened more severely over time compared to L-cysteine hydrochloride (L-CH). Concomitant with dough hardening over time, the decline in free thiols and two-stage oxidation kinetics confirmed disulfide reformation. Raman spectroscopy indicated the formation of more stable disulfide configuration (gauche-gauche-gauche) in glutenin. Polymerization of proteins larger than 80 kDa was promoted, and extractability of high-molecular-weight (HMW) and B/C-low-molecular-weight (LMW) glutenin subunits (GS) was reduced. Disulfide cleavage reduced glutenin alpha-helix, while reformation increased beta-sheet/alpha-helix in L-CH but decreased beta-sheet in SMBS systems. Fluorescence intensity decreased in glutenin. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) identified HMW-GS PW212 (Cys46, Cys31), Dx5 (Cys118), Dy10 (Cys636) and LMW-GS 1D1 (Cys25) as key participants in disulfide dynamics, revealing an irreversible reformation process that preferentially establishes new intermolecular rather than original intrachain bonds.
Digital light processing (DLP) has been used to fabricate alumina (Al2O3) ceramics in order to evaluate their reciprocating wear behaviour. Although DLP can produce high-resolution parts with visually smooth surfaces, microscopic examination reveals staircase effects arising from the layer-by-layer AM process, which can influence tribological performance. Dry reciprocating wear tests were performed using a silicon nitride (β-Si3N4) counter face sphere under applied normal loads varied from 20 to 60 N. These conditions were determined from Hertzian contact mechanics and ceramic wear maps to capture transitions from mild to severe wear, with loads exceeding 60 N are known to induce ultra-severe wear regimes. Two specimen sets were investigated: (1) DLP builds with a constant 25 μm layer thickness, printed at orientations of 0° to 90° (relative to the build plate surface) in 15° increments, and (2) 45° inclined builds fabricated with layer thicknesses of 10, 25, 35, 50, and 75 μm; note that 35 μm is nominally equivalent to the pixel size for the printer used in this work. Increasing layer thickness reduces print time but degraded surface definition, while the combined effects of layer height and build orientation produced periodic surface features (the so-called ’staircase effect’) that strongly governed wear mechanisms, which were identified as deformation, localized microfracture, and micro-abrasion. The measured wear rate ranged from 10−5 to 10−6 mm3/N·m, and these findings highlight the interplay between DLP process parameters, surface topology, and the tribological response in additively manufactured alumina ceramics, providing valuable information relating to component design.
