114333 research outputs found
Sort by
Biochar mitigates N2O and NH3 emissions in sheep manure composting by regulating microbial genes associated with nitrogen cycle
No full textImproper treatment of livestock and poultry manure from animal husbandry could lead to substantial Ammonia (NH3) and Nitrous oxide (N2O) emissions. In this study, rice husk biochar (RHB) and sawdust biochar (SDB) were investigated as additives to control NH3 and N2O emissions during sheep manure composting. Metagenomic sequencing was conducted to characterize the microbiome during composting and functional genes associated with nitrogen cycle (NC genes), and correlations among properties of composting feedstocks, NH3 and N2O emissions, NC genes were investigated. The results showed that RHB and SDB significantly reduced both NH3 emission (42.0 % and 61.3 %, respectively) and N2O emission (17.5 %, and 41.7 %, respectively), yet the reduction exhibited only a weak correlation with the physical attributes of the RHB. Microbial co-occurrence network showed biochar enhanced the complexity of the microbial community and the positive-to-negative correlation ratios. RDA analysis elucidated that biochar additives improved both total organic carbon (TOC) and C/N by inhibiting denitrification-related narG/Z/nxrA, nirK and norB/C genes while promoting nosZ, and inhibiting mineralization-related GLUD1_2/gdhA, E3.5.1.49, E3.5.5.1, gudB/rocG genes. PLS-PM model further confirmed that TOC was negatively correlated to NC genes involved in N2O- and NH3- production. The study provided theoretical basis and a road map to develop efficient composting additives to regulate TOC and C/N and to reduce NH3 and N2O emissions.This article is published as Wang, Yi, Xinyao Fan, Wenqi Liang, Xu Ma, Wenming Zhang, and Chenxu Yu. "Biochar mitigates N2O and NH3 emissions in sheep manure composting by regulating microbial genes associated with nitrogen cycle." Environmental Technology & Innovation (2026): 104828. doi: https://doi.org/10.1016/j.eti.2026.104828.This work was supported by Major Special Research Projects in Gansu Province [25ZDNA007]
The effect of geometry on the microstructure, crystallographic texture, and mechanical properties of electron beam powder bed fusion additively manufactured Haynes 282 superalloy
No full textDuring the additive manufacturing of metals and alloys, including Haynes 282 that has been processed using electron beam powder bed fusion (E-PBF), the local geometry, particularly near large intentionally engineered internal features, free surfaces, and thin walls, can significantly modify thermal gradients, grain morphology, and texture. Thermal simulations reveal that the designed pores act as thermal insulators, disrupting cooling rates and promoting heterogeneous nucleation and grain growth competition near the top surface of the pores. Twinning was observed within the equiaxed grains near designed pores and sample edges, hinting at grain morphology, orientation, and size as key variables in the twinning mechanism. Thin-walled struts on the top of the samples help analyze geometry-dependent shifts in heat extraction and grain growth, resulting in distinct angled columnar grains.
Detwinning in Haynes 282 (H282) was investigated using thermomechanical reversal experiments that partly replicate thermal cycle conditions of additive manufacturing. Twinning was effectively eliminated through Gleeble simulations conducted between 1000–1250°C. While lower-temperature cycling (300–800°C) caused only cyclic hardening, high-temperature cycles promoted twin boundary annihilation, grain growth, and dynamic recovery. EBSD and KAM analyses confirmed a significant reduction of Σ3 twin boundaries and increased disorientation activity near grain boundaries, indicating strain-induced lattice reorientation. Overall, elevated-temperature thermomechanical gyrations provide an effective route for twin elimination and microstructural homogenization in nickel-based superalloys.
The influence of part geometry on crystallographic texture evolution in electron beam powder bed fusion (E-PBF)–fabricated Haynes 282 (H282) superalloy was investigated. Electron backscatter diffraction (EBSD) was used to characterize texture variations in regions containing engineered cubic and spiral pores. The results revealed a strong [001] texture in the bulk region, corresponding to columnar grain growth along the build direction, and weak or random textures near internal pores where thermal gradients were disrupted. The spiral pore specimen demonstrated a gradual texture transition with increasing height above the pore, where multiple of uniform density (MUD) values increased from ~2.7 to ~11, reflecting recovery of directional solidification. A new analytical framework introducing an additional Euler rotation (θ) was developed to correct for out-of-plane EBSD acquisition and validated using a synthetic orientation dataset. The new reference frame transformation enabled accurate texture mapping in inclined geometries, confirming that geometric features act as localized thermal anomalies governing solidification mode, grain morphology, and anisotropy. Overall, the findings establish a clear link between geometry-driven heat flow variations and texture heterogeneity in E-PBF H282, with implications for tailoring microstructure and mechanical performance in complex additive-manufactured components.
A probabilistic, data-driven framework was developed to predict local mechanical properties in Electron Beam Melted (EBM) Haynes 282 by integrating tensile, nano-, and micro-indentation datasets from conventionally processed (hot-rolled, heat-treated, Gleeble) and geometry-resolved EBM regions (bulk, above-pore, and 2, 3, & 4 mm struts). Cumulative distribution functions quantify spatial variability in hardness arising from geometry-dependent thermal histories; the bulk exhibits the highest nanohardness (~4.63 GPa) and yield strength (537 MPa), while struts, especially 3 mm, shift toward lower hardness. A multivariate regression model links yield strength to nano- and micro-hardness with high fidelity (R² > 0.92, RMSE ~ 0.029 GPa), and estimates microhardness directly from nanohardness. Distribution Translation and Rotation (DTR) calibration aligns predicted and measured microhardness CDFs across regions, preserving both medians and spreads; deviations are typically < 0.05 GPa (e.g., bulk: 3.21–3.23 GPa; 2 mm strut: 3.17–3.18 GPa). The results establish indentation-based, distribution-calibrated modeling as an efficient route for site-specific property mapping in complex AM geometries, enabling reliable estimation of yield strength and hardness where direct testing is impractical
MaizeField3D: A curated 3D point cloud and procedural model dataset of field-grown maize from a diversity panel
No full textThe development of artificial intelligence (AI) and machine learning (ML) based tools for 3D phenotyping, especially for maize, has been limited due to the lack of large and diverse 3D datasets. 2D image datasets fail to capture essential structural details such as leaf architecture, plant volume, and spatial arrangements that 3D data provide. To address this limitation, we present MaizeField3D (\href{https://baskargroup.github.io/MaizeField3D/}{website}), a curated dataset of 3D point clouds of field-grown maize plants from a diverse genetic panel, designed to be AI-ready for advancing agricultural research. Our dataset includes 1,045 high-quality point clouds of field-grown maize collected using a terrestrial laser scanner (TLS). Point clouds of 520 plants from this dataset were segmented and annotated using a graph-based segmentation method to isolate individual leaves and stalks, ensuring consistent labeling across all samples. This labeled data was then used for fitting procedural models that provide a structured parametric representation of the maize plants. The leaves of the maize plants in the procedural models are represented using Non-Uniform Rational B-Spline (NURBS) surfaces that were generated using a two-step optimization process combining gradient-free and gradient-based methods. We conducted rigorous manual quality control on all datasets, correcting errors in segmentation, ensuring accurate leaf ordering, and validating metadata annotations. The dataset also includes metadata detailing plant morphology and quality, alongside multi-resolution subsampled point cloud data (100k, 50k, 10k points), which can be readily used for different downstream computational tasks. MaizeField3D will serve as a comprehensive foundational dataset for AI-driven phenotyping, plant structural analysis, and 3D applications in agricultural research
Interpretable time-series forecasting with multi-model deep learning and natural language processing (NLP) driven explainable artificial intelligence (XAI)
No full textDespite the remarkable predictive power of deep learning models, their "black-box" nature remains a critical barrier to adoption in high-stakes domains such as healthcare and industrial systems. This thesis addresses the challenge of achieving both high accuracy and model interpretability in multivariate temporal regression tasks by proposing a hybrid framework that integrates deep learning architectures (e.g., ResNet, Transformer), Explainable AI (XAI), and Natural Language Processing (NLP).
The research reimagines sensor data streams as 2D heatmaps, enabling the use of image-processing techniques to extract spatial-temporal patterns. By normalizing heterogeneous sensor units to a unified [0,1] range and organizing time-series data into grayscale images, the framework leverages Convolutional Neural Networks (CNNs) and self-attention mechanisms to capture localized anomalies and global dependencies. A ResNet-Transformer fusion architecture is introduced, combining ResNet’s ability to detect fine-grained sensor-specific features with the Transformer’s capacity to model long-range temporal correlations. This hybrid approach achieves state-of-the-art performance on two diverse datasets: the UCI Energy Appliance dataset (20% reduction in RMSE compared to standalone models) and an ECG arrhythmia detection task (3.8% accuracy improvement over InceptionTime).
To address interpretability, the framework incorporates XAI techniques such as Grad-CAM and SHAP. These methods generate sensor-level explanations by highlighting critical regions in input heatmaps and quantifying feature contributions. For instance, Grad-CAM identifies clinically relevant ECG segments (e.g., prolonged QT intervals), while SHAP values validate sensor pruning strategies by flagging low-impact sensors (e.g., those with variance below 0.1%). The thesis also explores the robustness of these explanations through deletion tests and sensitivity analysis, demonstrating that masking the top 20% of heatmap regions degrades accuracy by 41.3% (versus 28.5% for ResNet alone), underscoring the framework’s faithfulness to input data.
A novel contribution lies in automating the generation of human-readable explanations via NLP. By mapping attention weights and SHAP scores to domain-specific terminology (e.g., “elevated ST-segment suggests myocardial ischemia”), the system produces actionable insights for non-experts. BLEU-4 and ROUGE-L scores (0.586 and 0.65, respectively) validate the linguistic quality of these explanations, while domain expert evaluations rate clarity at 4.7 out of 5.
The thesis further advances data preprocessing by formalizing a min-max normalization pipeline that preserves temporal patterns while harmonizing heterogeneous sensor units. This approach reduces computational overhead by 40% through sensor pruning and timestamp downsampling, validated on the UCI Energy dataset.
Key findings include:
1. Superior Predictive Accuracy: The hybrid model outperforms standalone baselines (ResNet, Transformer, LSTM) by 3.2–4.5% in F1-score.
2. Interpretable Design: Grad-CAM heatmaps align with clinical intuition (e.g., ECG ST-segment elevation), while SHAP values enable data-driven feature selection.
3. Real-World Impact: Applications to energy consumption forecasting and cardiac arrhythmia detection demonstrate practical utility, with the framework reducing false negatives in medical diagnostics by 12.4% and improving grid efficiency in energy systems by 15%.
This work bridges the gap between deep learning performance and transparency, offering a blueprint for deploying AI in regulated environments. Future directions include dynamic attention mechanisms and automated sensor pruning pipelines for edge computing. By unifying pattern analysis, multi-model fusion, and NLP-driven explanations, the thesis contributes a scalable, interpretable framework for time-series regression in critical domains
Defining and detecting bone mineral deficiency in swine: From visual assessment to predictive modeling and practical alternatives to dual energy x-ray absorptiometry
No full textPhosphorus (P) underpins skeletal integrity, locomotor function, and overall welfare in swine production, yet current field diagnostics seldom detect dietary P inadequacy before performance losses occur. This dissertation integrates phenotypic, imaging, and chemical approaches to build a diagnostic pipeline for identifying P deficiency in growing pigs. Dietary‐response phenotyping showed that locomotion score (LS) is a sensitive, on-farm indicator of P status during the grower period. Mixed-sex pigs were fed available-phosphorus–adequate (ADQ), marginal (MARG), or deficient (DEF) diets across nursery, grower, and finisher phases. DEF diets produced sharp reductions in femur bone mineral content (FBMC), femur bone mineral density (FBMD), and LS only in grower pigs; a 14-d repletion diet restored LS, while nursery and finisher pigs were unaffected. These data identify 20–40 kg body weight as the most P-sensitive window and validate LS as a practical surveillance tool. To establish quantitative bone-health benchmarks, dual-energy X-ray absorptiometry (DXA) was coupled with generalized additive models (GAMs). This work evaluated bone characteristics for pigs fed diets supplemented with mono-calcium phosphate (MCP) as the sole source of P and pigs fed diets containing microbial phytase (Phytase) to increase bioavailability from phytate bound plant sources. In pigs fed MCP, whole-body (WB) BMD was predicted from
body weight with high precision (Adj R² = 0.94) and classified deficient pigs with 93.4% accuracy during the grower phase; models for Phytase were less robust, reflecting variable P release and altered Ca:P ratios. The resulting body-weight-specific reference intervals constitute the first empirically derived definition of “normal” BMD in adequately nourished pigs.
A low-cost alternative to DXA was evaluated by modeling proximate bone-ash traits. Seventy-three pigs consuming diets with graded Ca:P ratios were scanned by DXA and analyzed chemically. Femur ash BMD and BMC predicted DXA outcomes with Adj R² > 0.97, whereas third-metatarsal (3 MT) ash had poor diagnostic value; classification of deficiency peaked at 55% for WB-DXA BMD. Thus, femur ash is a practical surrogate where imaging is unavailable, especially in growing pigs, although deficiency classification remains more reliable with DXA. Collectively, these studies deliver a hierarchical diagnostic pipeline: (i) LS enables rapid field screening during the grower phase; (ii) DXA-based GAMs provide weight-adjusted thresholds for bone health and precise deficiency classification; and (iii) femur ash models
extend diagnostic capacity to settings lacking imaging equipment. By converging phenotypic, imaging, and chemical assessments, this work advances precision phosphorus nutrition and equips producers with actionable tools to safeguard bone health, welfare, and productivity in commercial swine systems
"Characterizing, modelling, and simulation of the agricultural rubber track system "
No full textThis dissertation presents a investigation into dynamic modeling, simulation, and experimental characterization of rubber track systems used in agricultural machinery. The primary objectives of this study are threefold: (1) to investigate and quantify the material properties of rubber track composites through carefully designed physical experiments, focusing on stiffness and damping characteristics; (2) to develop a two-dimensional multibody dynamic model of a continuous rubber track system based on first principles and calibrated using experimentally identified system parameters; and (3) to evaluate model through both qualitative and quantitative analyses of its dynamic responses, with sensitivity analysis performed to assess the influence of key parameters.
The first objective of this study was to investigate and quantify the material properties of rubber track composites through carefully designed physical experiments. Tensile and bending tests were performed on fiber- and steel-reinforced track specimens using an Instron 3367 universal testing machine with custom-designed clamping fixtures to ensure secure gripping and controlled loading. The experimental design included multiple strain rates (5–500 mm/min) to capture potential viscoelastic rate effects, with each condition repeated to assess consistency. Data were analyzed using linear mixed-effects models (LMMs) to statistically separate fixed effects (displacement, strain rate, bending angle) from random effects (replications, experimental order). This provided unbiased estimates of stiffness and damping parameters while quantifying variability across tests. A physics-based scaling framework, combining series–parallel spring analogies with geometric similarity principles, was then applied to extrapolate specimen-level measurements to full-scale track properties. This approach yielded global linear stiffness, damping, and rotational parameters suitable for use in system-level modeling.
As the second objective, a two-dimensional multibody dynamics (MBD) model of a continuous rubber track was developed from first principles. The belt is discretized into rigid segments; adjacent segments are joined by torsional spring–damper elements calibrated from the coupon tests, representing the track’s distributed stiffness and energy dissipation. Track–wheel–ground interactions are written using normal and tangential contact laws. The experimentally derived parameters are embedded in a geometrically simplified yet physically faithful MATLAB implementation that resolves the coupled dynamics among the drive sprocket, idlers, tensioner, and the compliant belt. The model reproduces both steady-state and transient responses under realistic loading and boundary conditions, enabling calibrated simulations for subsequent analyses.
The third objective was to evaluate the predictive performance of the MBD model through qualitative and quantitative analyses of its dynamic responses. Simulated chassis and wheel velocities and their frequency content were analyzed, and robustness was assessed via sensitivity studies that varied linear stiffness, linear damping, rotational stiffness, rotational damping, and segment count by ±10% about nominal values. Among these factors, rotational stiffness and rotational damping exerted the greatest influence on key metrics (RMS chassis velocity, dominant frequency, and tractive efficiency), indicating that accurate identification of segment-level torsional properties is most critical. Under a driver-roller step from 3.14 to 6.28 rad/s (30 to 60 rev/min), the chassis velocity and idler angular velocity exhibited an underdamped, second-order transient, consistent with the model’s dominant dynamic modes. The model captured power transmission through the compliant belt by resolving deformation, internal damping, and contact interactions. Tractive efficiency increased with draft, peaking at 32% near 1800 N and then declining at higher loads, reflecting the nonlinear balance between useful traction and internal losses; simulations attribute the high-load efficiency drop primarily to bending losses as the belt repeatedly wraps the drive and idler
Silyl-mediated Deacetylation vs Heat-up Synthesis of Chalcohalides: Pushing the Size and Composition Envelope
No full textChalcohalide semiconductors are rapidly gaining traction as stable, biocompatible materials for energy conversion applications. While the solid-state synthesis of bulk chalcohalides is relatively well-developed, the colloidal chemistry of these materials is still in its early stages. Colloidal semiconductors are often advantageous in device fabrication due to the cost effectiveness of solution processing. Thus, we aim to increase the utility of chalcohalides in device fabrication by establishing solution phase chemistry of promising compositions. We show that silyl-mediated deacetylation is a versatile and effective method of making colloidal PnChI (Pn = Sb, Bi; Ch = S, Se) and Sn2PnS2I3 (Pn = Sb, Bi) chalcohalides of tunable sizes and compositions. Furthermore, we demonstrate the preparation of mixed-pnictide chalcohalides through silyl-mediated deacetylation and/or cation exchange, the latter being one of the few reported instances in chalcohalides. Additionally, we use the thiocyanate heat up approach in combination with density functional theory to study halide mixing in quaternary tin chalcohalides. By pushing the limits of each synthetic technique, we have designed more soluble chalcohalides with tunable compositions while also gaining a better understanding of the efficacy of each procedure in respect to thin film and subsequent device fabrication. In addition to size and composition tuning, silyl-mediated deacetylation can help facilitate the future development and wide-scale application of chalcohalide-based devices by expanding the selection of solution-processable chalcohalides.This is a preprint from Stegner, Eve, Md Riad Sarkar Pavel, Anuluxan Santhiran, Jack Lawton, Juan Pablo Correa-Baena, Aaron Rossini, and Javier Vela. "Silyl-mediated Deacetylation vs Heat-up Synthesis of Chalcohalides: Pushing the Size and Composition Envelope." (2026). doi: https://doi.org/10.26434/chemrxiv-2025-90rl2/v2.This work was supported by the U.S. National Science Foundation, Division of Chemistry, Macromolecular, Supramolecular, and Nanochemistry Program (2305062) by a grant awarded to J. V.. SSNMR work by A.S. and A.J.R. was supported by the U.S. Department of Energy (DOE) Ames National Laboratory, Materials Science and Engineering Division. Ames National Laboratory is operated for the U.S. DOE by Iowa State University, under contract no. DE-AC02- 07CH11358. We thank Allie Roth and Gordie Miller for comments
Prenatal and postnatal atrazine-induced endocrine and ovarian chemical biotransformation protein disruption in offspring
No full textAtrazine (ATZ) is an endocrine-disrupting chemical, and ATZ exposure during perinatal development is linked to reproductive dysfunction and behavioral abnormalities in adulthood. Gestational ATZ exposure can adversely affect birth weight, fetal growth, and onset of puberty. To investigate ovarian effects of ATZ exposure on female offspring, timed pregnant nulliparous gilts were provided ad libitum access to water that contained vehicle control (CT; 0.002% (v/v) ethanol; n = 3) or an environmentally relevant ATZ dose (20 µg/L; n = 4); thus piglets were exposed from gestation day 28 through farrowing and lactation (∼99 days). Umbilical cord blood serum was assayed for hormone concentrations, ovarian follicles were classified and counted, and abundance of proteins involved in folliculogenesis, steroidogenesis, chemical biotransformation, DNA damage, and cell viability in piglet ovaries were quantified by Western blotting. Exposure to ATZ decreased (P < .05) body weight at postnatal day 10, ovarian abundance of cytochrome P450 (CYP) isoform 2E1 (CYP2E1), ATP-binding cassette subfamily B member 1 (ABCB1), CYP11A1, peroxisome proliferator–activated receptor α, CYP1A1, CYP1B1, pAKT, RAD51, and serum progesterone. There was a tendency (.05 < P < .10) for reduced birth weight, serum 17β-estradiol, atretic follicle number, and ovarian GSTP1 and for increased PARP1 abundance after ATZ exposure. Overall, this study suggests that ATZ exposure during gestation and lactation may impair ovarian function.This article is published as Omotoyosi Adeyanju, Maria Estefanía González-Alvarez, Collins Antwi-Boasiako, Mary J Laws, Amy T Desaulniers, Aileen F Keating, Prenatal and postnatal atrazine-induced endocrine and ovarian chemical biotransformation protein disruption in offspring, Journal of the Endocrine Society, Volume 10, Issue 3, March 2026, bvag027, https://doi.org/10.1210/jendso/bvag027This work was supported by the University of Nebraska Collaboration Initiative and by Iowa State University Incentive funds
Docking in the Abies grandis abietaenol synthase to investigate the mechanism underlying conifer resin acid diterpene synthase evolution
No full textThe pine family (Pinaceae) produces oleoresins in which the diterpenoid resin acids contain tricyclic backbones of either the (iso)pimarane and/or rearranged abietane type. These are produced by closely related (class I) diterpene synthases, whose functional divergence is of significant interest, not least for mechanistic insights into the catalyzed complex carbocation-cascade reactions. The abietaenol synthase from Abies grandis (AgAS) has long served as a model for this enzymatic subfamily and was used here to examine key residues potentially involved in transitions from such ancestral activity to production of (iso)pimaradienes, which seems to have occurred independently at least three times in the Pinaceae. While the equivalent substitutions in AgAS led to production of some amount of isopimaradiene(s), based on the transition observed in the most closely related enzymes (i.e., those also from the Abies genus), the threonine substitution for an alanine was most impactful, with the resulting A723T variant producing almost entirely isopimara-7,15-diene. This prompted application of the TerDockin computational approach to investigate the underlying mechanism. Through iterative application, requiring placement and constraint of the reactant water from the native reaction (i.e., that added to yield the 13α-hydroxyl of the primary abietaenol epimer), it was found that the introduced threonine most likely acts directly as a catalytic base to short-circuit the native reaction by deprotonating the initially formed isopimara-13E-en-8-yl carbocation intermediate. These results not only provide mechanistic insight but also have some implications for such modeling of the complex reactions catalyzed by terpene synthases more generally, as discussed herein.This article is published as Mark Schmidt-Dannert, Griffin Humphreys, Taylor Eich, Meirong Jia, Reuben J. Peters; Docking in the Abies grandis abietaenol synthase to investigate the mechanism underlying conifer resin acid diterpene synthase evolution. Biochem J 4 March 2026; 483 (3): 365–374. doi: https://doi.org/10.1042/BCJ20250294This work was supported by a grant from the National Institutes of Health [GM156300 to R.J.P.] for the work done in the U.S.A. and a grant from the National Natural Science Foundation of China [22207129 to M.J.] for the work done in China
Retrospective Report of the Use and Outcome of High-Velocity Nasal Insufflation in Cats (2019–2022): Eight Cases
No full textObjective: To describe the use and feasibility of high-velocity nasal insufflation (HVNI) in cats. Design: Retrospective descriptive study from 2019 to 2022. Setting: University teaching hospital.
Animals: Eight cats that failed traditional oxygen therapy and, based on clinical evaluation, required more aggressive oxygen supplementation. Measurements and Main Results: Eight cats had HVNI instituted between 2019 and 2022. Four cats received HVNI for primary pulmonary disease, including two with severe nodular pulmonary patterns, one with acute respiratory distress syndrome secondary to septic shock, and one with diffuse hepatization of multiple lung lobes. Two cats received HVNI for airway disease, including one with feline asthma and one with chronic bronchiolar injury. One cat received HVNI due to pleural space disease secondary to failed herniorrhaphy repair for a peritoneal–pericardial diaphragmatic hernia, and one had an unknown cause of respiratory distress. No cat received HVNI due to cardiac disease. The median time spent on HVNI was 17 h (range, 2–76 h). The median flow rate was 1 L/kg (range, 0.612–2.68 L/kg). The median highest FiO2 recorded was 100% (range, 50%–100%), and the median lowest was 52.5% (range, 40%–100%). Three cats had HVNI successfully discontinued, and two cats survived to discharge. All cats tolerated HVNI with no reported complications. Conclusions: This is the first report to evaluate the use of HVNI for respiratory support in cats, demonstrating that HVNI is a feasible option in cats requiring more aggressive respiratory support than traditional oxygen therapy.This article is published as Wampfler, A., R. A. Walton, M. ‘t Hoen, J. Ward, and A. E. Blong. "Retrospective Report of the Use and Outcome of High‐Velocity Nasal Insufflation in Cats (2019–2022): Eight Cases." Journal of Veterinary Emergency and Critical Care. doi: https://doi.org/10.1111/vec.70085