Search NASA⌕ Search

SEARCH · Search NASA

Results for “Predictive modeling”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 973 records · Page 54

Mitigating Errors in External Respiratory Surrogate-Based Models of Tumor Position

Purpose: To investigate the effect of tumor site, measurement precision, tumor-surrogate correlation, training data selection, model design, and interpatient and interfraction variations on the accuracy of external marker-based models of tumor position. Methods and Materials: Cyberknife Synchrony system log files comprising synchronously acquired positions of external markers and the tumor from 167 treatment fractions were analyzed. The accuracy of Synchrony, ordinary-least-squares regression, and partial-least-squares regression models for predicting the tumor position from the external markers was evaluated. The quantity and timing of the data used to build the predictive model were varied. The effects of tumor-surrogate correlation and the precision in both the tumor and the external surrogate position measurements were explored by adding noise to the data. Results: The tumor position prediction errors increased during the duration of a fraction. Increasing the training data quantities did not always lead to more accurate models. Adding uncorrelated noise to the external marker-based inputs degraded the tumor-surrogate correlation models by 16% for partial-least-squares and 57% for ordinary-least-squares. External marker and tumor position measurement errors led to tumor position prediction changes 0.3-3.6 times the magnitude of the measurement errors, varying widely with model algorithm. The tumor position prediction errors were significantly associated with the patient index but not with the fraction index or tumor site. Partial-least-squares was as accurate as Synchrony and more accurate than ordinary-least-squares. Conclusions: The accuracy of surrogate-based inferential models of tumor position was affected by all the investigated factors, except for the tumor site and fraction index.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Machine Learning Predictions of a Multiresolution Climate Model Ensemble

Statistical models of high-resolution climate models are useful for many purposes, including sensitivity and uncertainty analyses, but building them can be computationally prohibitive. We generated a unique multiresolution perturbed parameter ensemble of a global climate model. We use a novel application of a machine learning technique known as random forests to train a statistical model on the ensemble to make high-resolution model predictions of two important quantities: global mean top-of-atmosphere energy flux and precipitation. The random forests leverage cheaper low-resolution simulations, greatly reducing the number of high-resolution simulations required to train the statistical model. We demonstrate that high-resolution predictions of these quantities can be obtained by training on an ensemble that includes only a small number of high-resolution simulations. Finally, we also find that global annually averaged precipitation is more sensitive to resolution changes than to any of the model parameters considered.

54 ENVIRONMENTAL SCIENCES↗

Pressure-Gradient-Based RANS Model for Predicting Separation in Transitional and Turbulent Flows

Predicting flow separation poses a significant challenge for RANS models, particularly in transitional flows over airfoils. We propose a novel improvement to RANS models to predict incipient separation in both transitional and fully turbulent flows. Our approach modifies the eddy viscosity model in regions indicated by a pressure-gradient criterion that accounts for intermittency - determining whether the boundary layer is laminar or turbulent. This model demonstrates robust generalization across various airfoil shapes and Reynolds numbers. Applied to the NREL Phase VI wind turbine rotor, our model shows improved aerodynamic performance predictions compared to the baseline RANS model.

k-omega SST↗

A Fast, Accurate Prediction for System-Wide Damage Due to Dynamic Wind Loading

The complex relationship between photovoltaic (PV) hardware configurations, overall system dynamics, and turbulent aerodynamic phenomena generates highly unsteady, non-uniform loads that can lead to damaging instabilities. These effects may result in glass breakage, cell cracking, and structural failures in frames and mounting systems, even under moderate wind conditions. Addressing industry concerns about premature system failures in field conditions deemed survivable, our research aims to develop a fast and accurate predictive model for system damage. This model integrates configurable hardware choices with advanced simulation tools to represent the overall system-specific dynamics effectively. Using this model, we predict responses under varying weather conditions and hardware setups, translating these predictions into pre-trained surrogate models capable of accurately identifying failure risks and rapidly testing new system hardening measures. In this presentation, we will showcase preliminary results in capturing system dynamics through our customizable library of PV hardware configurations. Additionally, we will highlight how these new tools build upon PVade's established wind load modeling capabilities and foster the development of advanced AI/ML surrogates for improving system robustness.

97 MATHEMATICS AND COMPUTING↗

An L$_0$L$_1$-norm compressive sensing paradigm for the construction of sparse predictive lattice models using mixed integer quadratic programming

First-principles based lattice models allow the modeling of ab initio thermodynamics of crystalline mixtures for applications such as the construction of phase diagrams and the identification of ground state atomic orderings. The recent development of compressive sensing approaches for the construction of lattice models has further enabled the systematic construction of sparse physical models without the need for human intuition other than requiring the compactness of effective cluster interactions. However, conventional compressive sensing based on L1-norm regularization is strictly only applicable to certain classes of optimization problems and is otherwise not guaranteed to generate optimally sparse and transferable results, so that the method can only be applied to some materials science applications. In this paper, we illustrate a more robust L0L1-norm compressive-sensing method that removes the limitations of conventional compressive sensing and generally results in sparser lattice models that are at least as predictive as those obtained from L1-norm compressive sensing. Apart from the theory, a practical implementation based on state-of-the-art mixed-integer quadratic programming (MIQP) is proposed. The robustness of our methodology is illustrated for four different transition-metal oxides with relevance as battery cathode materials: Li2xTi2(1-x)O2, Li2xNi2yO2, MgxCr2O4, and NaxCrO2. This method provides a practical and robust approach for the construction of sparser and more predictive lattice models, improving on the compressive sensing paradigm and making it applicable to a much broader range of applications.

Huang, W↗

Evaluation of the 239 Pu prompt fission neutron spectrum induced by neutrons of 500 keV and associated covariances

We present evaluations of the prompt fission neutron spectrum (PFNS) of (PU)-P-239 induced by 500 keV neutrons, and associated covariances. In a previous evaluation by Talon et al. (2010), surprisingly low evaluated uncertainties were obtained, partly due to simplifying assumptions in the quantification of uncertainties from experiment and model. Therefore, special emphasis is placed here on a thorough uncertainty quantification of experimental data and of the Los Alamos model predicted values entering the evaluation. In addition, the Los Alamos model was extended and an evaluation technique was employed that takes into account the qualitative differences between normalized model predicted values and experimental shape data These improvements lead to changes in the evaluated PENS and overall larger evaluated uncertainties than in the previous work. However, these evaluated uncertainties are still smaller than those obtained in a statistical analysis using experimental information only, due to strong model correlations. Hence, suggestions to estimate model defect uncertainties are presented. which lead to more reasonable evaluated uncertainties. The calculated k(eff) of selected criticality benchmarks obtained with these new evaluations agree with each other within their uncertainties despite the different approaches to estimate model defect uncertainties. The k(eff) one standard deviations overlap with some of those obtained using ENDF/B-VILl, albeit their mean values are further away from unity. Spectral indexes for the Jezebel critical assembly calculated with the newly evaluated PFNS agree with the experimental data for selected (n,) and (n,f) reactions, and show improvements for highenergy threshold (n,2n) reactions compared to ENDF/B-VII.l. (C) 2015 Elsevier B.V. All rights reserved.

Neudecker, D.↗

Investigating calcite growth rates using a quartz crystal microbalance with dissipation (QCM-D)

Calcite precipitation plays a significant role in processes such as geological carbon sequestration and toxic metal sequestration and, yet, the rates and mechanisms of calcite growth under close to equilibrium conditions are far from well understood. In this study, a quartz crystal microbalance with dissipation (QCM-D) was used for the first time to measure macroscopic calcite growth rates. Calcite seed crystals were first nucleated and grown on sensors, then growth rates of calcite seed crystals were measured in real-time under close to equilibrium conditions (saturation index, SI = log ({Ca 2+ }/{CO 3 2– }/ K sp ) = 0.01–0.7, where {i} represent ion activities and K sp = 10 –8.48 is the calcite thermodynamic solubility constant). At the end of the experiments, total masses of calcite crystals on sensors measured by QCM-D and inductively coupled plasma mass spectrometry (ICP-MS) were consistent, validating the QCM-D measurements. Calcite growth rates measured by QCM-D were compared with reported macroscopic growth rates measured with auto-titration, ICP-MS, and microbalance. Calcite growth rates measured by QCM-D were also compared with microscopic growth rates measured by atomic force microscopy (AFM) and with rates predicted by two process-based crystal growth models. The discrepancies in growth rates among AFM measurements and model predictions appear to mainly arise from differences in step densities, and the step velocities were consistent among the AFM measurements as well as with both model predictions. Using the predicted steady-state step velocity and the measured step densities, both models predict well the growth rates measured using QCM-D and AFM. Furthermore, this study provides valuable insights into the effects of reactive site densities on calcite growth rate, which may help design future growth models to predict transient-state step densities.

58 GEOSCIENCES↗

The impact of including spatially longitudinal heterogeneities of vessel oxygen content and vascular fraction in 3D tumor oxygenation models on predicted radiation sensitivity

Purpose: Oxygen distribution models have been used to analyze the influences of oxygen tensions on tissue response after radiotherapy. These distributions are often generated assuming constant oxygen tension in the blood vessels. However, as red blood cells progress through the vessels, oxygen is continuously released into the plasma and the surrounding tissue, resulting in longitudinally varying oxygen levels in the blood vessels. In the present study, the authors investigated whether a tumor oxygenation model that incorporated longitudinally varying oxygen levels would provide different predictions of necrotic fractions and radiosensitivity compared to commonly used models with a constant oxygen pressure. Methods: Our models simulated oxygen diffusion based on a Green's function approach and oxygen consumption according to the Michaelis-Menten equation. The authors constructed tumor models with different vascular fractions (VFs), from which they generated depth oxygenation curves and a look-up table of oxygen pressure gradients. The authors evaluated models of spherical tumors of various sizes, from 1 to 10{sup 4} mg. The authors compared the results from a model with constant vessel oxygen (CVO) pressure to those from models with longitudinal variations in oxygen saturation and either a constant VF (CVF) or variable VF (VVF) within the tumor tissue. The authors monitored the necrotic fractions, defined as tumor regions with an oxygen pressure below 1 mmHg. Tumor radiation sensitivity was expressed as D{sub 99,} the homogeneous radiation dose required for a tumor control probability of 0.99. Results: In the CVO saturation model, no necrosis was observed, and decreasing the VF could only decrease the D{sub 99} by up to 10%. Furthermore, the D{sub 99} vs VF dependence was similar for different tumor masses. Compared to the CVO model, the extended CVF and VVF models provided clearly different results, including pronounced effects of VF and tumor size on the necrotic fraction and D{sub 99}, necrotic fractions ranging from 0% to 97%, and a maximal D{sub 99} increment of 57%. Only minor differences were observed between different vessel architectures, i.e., CVF vs VVF. In the smallest tumor with a low necrotic fraction, the D{sub 99} strictly decreased with increasing blood velocity. Increasing blood velocity also decreased the necrotic fraction in all tumor sizes. VF had the most profound influence on both the necrotic fraction and on D{sub 99}. Conclusions: Our present analysis of necrotic formation and the impact of tumor oxygenation on D{sub 99} demonstrated the importance of including longitudinal variations in vessel oxygen content in tumor models. For small tumors, radiosensitivity was particularly dependent on VF and slightly dependent on the blood velocity and vessel arrangement. These dependences decreased with increasing tumor size, because the necrotic fraction also increased, thereby decreasing the number of viable tumor cells that required sterilization. The authors anticipate that the present model will be useful for estimating tumor oxygenation and radiation response in future detailed studies.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Mapping the proteogenomic landscape enables prediction of drug response in acute myeloid leukemia

Acute myeloid leukemia is a poor prognosis cancer commonly stratified by genetic aberrations, but these mutations are often heterogeneous and don’t always predict therapeutic response. Here we combine transcriptomic, proteomic, and phosphoproteomic datasets with ex vivo drug sensitivity data to help understand the underlying pathophysiology of AML beyond mutations. We measured the proteome and phosphoproteome of 210 patients and combined them with genomics and transcriptomic measurements to identify four proteogenomic subtypes that complemented existing genetic subtypes. We then built a predictor to classify samples into subtypes based on 147 molecular features and mapped them to a ‘landscape’. Each region of this landscape corresponded to specific drug response patterns. We then built a drug response prediction model to identify drugs that target distinct subtypes. We can ultimately use these models to predict drug treatment response and prioritize treatments. Finally, we extended our models and mapped a series of cell lines representing various stages of quizartinib resistance into our subtype landscape, predicting and experimentally validating a switch in sensitivity to venetoclax to panobinostat, two drugs with very different mechanisms than quizartinib. Our results show how multi-omics data together with drug sensitivity data can inform therapy stratification and drug combinations in AML.

59 BASIC BIOLOGICAL SCIENCES↗

A Comprehensive Investigation of Active Learning Strategies for Conducting Anti-Cancer Drug Screening

It is well-known that cancers of the same histology type can respond differently to a treatment. Thus, computational drug response prediction is of paramount importance for both preclinical drug screening studies and clinical treatment design. To build drug response prediction models, treatment response data need to be generated through screening experiments and used as input to train the prediction models. In this study, we investigate various active learning strategies of selecting experiments to generate response data for the purposes of (1) improving the performance of drug response prediction models built on the data and (2) identifying effective treatments. Here, we focus on constructing drug-specific response prediction models for cancer cell lines. Various approaches have been designed and applied to select cell lines for screening, including a random, greedy, uncertainty, diversity, combination of greedy and uncertainty, sampling-based hybrid, and iteration-based hybrid approach. All of these approaches are evaluated and compared using two criteria: (1) the number of identified hits that are selected experiments validated to be responsive, and (2) the performance of the response prediction model trained on the data of selected experiments. The analysis was conducted for 57 drugs and the results show a significant improvement on identifying hits using active learning approaches compared with the random and greedy sampling method. Active learning approaches also show an improvement on response prediction performance for some of the drugs and analysis runs compared with the greedy sampling method.

60 APPLIED LIFE SCIENCES↗

ACTIVE GALACTIC NUCLEI AS MAIN CONTRIBUTORS TO THE ULTRAVIOLET IONIZING EMISSIVITY AT HIGH REDSHIFTS: PREDICTIONS FROM A {Lambda}-CDM MODEL WITH LINKED AGN/GALAXY EVOLUTION

We have evaluated the contribution of the active galactic nuclei (AGN) population to the ionization history of the universe based on a semi-analytic model of galaxy formation and evolution in the cold dark matter cosmological scenario. The model connects the growth of black holes and of the ensuing AGN activity to galaxy interactions. In the model we have included a self-consistent physical description of the escape of ionizing UV photons; this is based on the blast-wave model for the AGN feedback we developed in a previous paper to explain the distribution of hydrogen column densities in AGNs of various redshifts and luminosities, due to absorption by the host galaxy gas. The model predicts UV luminosity functions for AGNs that are in good agreement with those derived from the observations especially at low and intermediate redshifts (z {approx} 3). At higher redshifts (z > 5), the model tends to overestimate the data at faint luminosities. Critical biases in both the data and in the model are discussed to explain such apparent discrepancies. The predicted hydrogen photoionization rate as a function of redshift is found to be consistent with that derived from the observations. All of the above suggests that we should reconsider the role of the AGNs as the main driver of the ionization history of the universe.

79 ASTRONOMY AND ASTROPHYSICS↗

Charge transport model to predict intrinsic reliability for dielectric materials

Several lifetime models, mostly empirical in nature, are used to predict reliability for low-k dielectrics used in integrated circuits. There is a dispute over which model provides the most accurate prediction for device lifetime at operating conditions. As a result, there is a need to transition from the use of these largely empirical models to one built entirely on theory. Therefore, a charge transport model was developed to predict the device lifetime of low-k interconnect systems. The model is based on electron transport and donor-type defect formation. Breakdown occurs when a critical defect concentration accumulates, resulting in electron tunneling and the emptying of positively charged traps. The enhanced local electric field lowers the barrier for electron injection into the dielectric, causing a positive feedforward failure. The charge transport model is able to replicate experimental I-V and I-t curves, capturing the current decay at early stress times and the rapid current increase at failure. The model is based on field-driven and current-driven failure mechanisms and uses a minimal number of parameters. All the parameters have some theoretical basis or have been measured experimentally and are not directly used to fit the slope of the time-to-failure versus applied field curve. Despite this simplicity, the model is able to accurately predict device lifetime for three different sources of experimental data. The simulation's predictions at low fields and very long lifetimes show that the use of a single empirical model can lead to inaccuracies in device reliability.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Remote sensing data assimilation for a prognostic phenology model

Predicting the global carbon and water cycle requires a realistic representation of vegetation phenology in climate models. However most prognostic phenology models are not yet suited for global applications, and diagnostic satellite data can be uncertain and lack predictive power. We present a framework for data assimilation of Fraction of Photosynthetically Active Radiation absorbed by vegetation (FPAR) and Leaf Area Index (LAI) from the MODerate Resolution Imaging Spectroradiometer (MODIS) to constrain empirical temperature, light, moisture and structural vegetation parameters of a prognostic phenology model. We find that data assimilation better constrains structural vegetation parameters than climate control parameters. Improvements are largest for drought-deciduous ecosystems where correlation of predicted versus satellite-observed FPAR and LAI increases from negative to 0.7-0.8. Data assimilation effectively overcomes the cloud- and aerosol-related deficiencies of satellite data sets in tropical areas. Validation with a 49-year-long phenology data set reveals that the temperature-driven start of season (SOS) is light limited in warm years. The model has substantial skill (R = 0.73) to reproduce SOS inter-annual and decadal variability. Predicted SOS shows a higher inter-annual variability with a negative bias of 5-20 days compared to species-level SOS. It is however accurate to within 1-2 days compared to SOS derived from net ecosystem exchange (NEE) measurements at a FLUXNET tower. The model only has weak skill to predict end of season (EOS). Use of remote sensing data assimilation for phenology model development is encouraged but validation should be extended with phenology data sets covering mediterranean, tropical and arctic ecosystems.

54 ENVIRONMENTAL SCIENCES↗

A physics-informed and hierarchically regularized data-driven model for predicting fluid flow through porous media

This paper presents a new deep learning data-driven model for predicting structure dependent pore-fluid velocity fields in rock. The model is based on a Convolutional Auto-Encoder (CAE) artificial neural network capable of learning from image data generated by direct numerical simulations of fluid flow through pore-structures, such as by Lattice Boltzmann or molecular dynamics methods. The main novelty of the model in comparison to previous CAE-based data-driven approaches consists of three parts. The first is a methodology for decomposing the full-domain of the porous media into sub-regions, or “sub-domains”, in order to reduce the overall size of the CAE, batch process the sub-domains in parallel, and enable the CAE to learn local and generalizable nonlinear mappings of pore-fluid velocities. The second consists of embedding the finite difference solutions of the incompressible Navier-Stokes and continuity equations into convolutional layers prior to the CAE in order to provide the CAE with knowledge of fluid dynamics physics (PhyFlow). The third main novelty is that the training of the CAE is regularized with a hierarchical loss function that encourages the learning of fluid flow patterns (in a way similar to ranked modes in principal component analysis), ranking from most to least important. This is shown to increase the stability in learning, reduce over-fitting, and promote interpretability of the CAE neural network layers (HierCAE). The comprehensive new data-driven model, which we call the PhyFlow-HierCAE model, is shown to exhibit improved accuracy and generalizability of flow field predictions over conventional CAE models, attributable to the embedded physical knowledge and the hierarchical regularization, as well as realize orders of magnitude speed-ups in computation times as a surrogate for the direct numerical simulations. Examples of training and forward predictions on unseen pore-structures are provided and evaluated for data from Lattice Boltzmann and molecular dynamics simulations of pore-fluid flow. The model is shown to be a fast and accurate emulator (or “surrogate”) for predicting effective permeability of unseen pore-structures based on learning from relatively small direct numerical simulation datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Percentage of Biopsy Cores Positive for Malignancy and Biochemical Failure Following Prostate Cancer Radiotherapy in 3,264 Men: Statistical Significance Without Predictive Performance

Purpose: To define and incorporate the impact of the percentage of positive biopsy cores (PPC) into a predictive model of prostate cancer radiotherapy biochemical outcome. Methods and Materials: The data of 3264 men with clinically localized prostate cancer treated with external beam radiotherapy at four institutions were retrospectively analyzed. Standard prognostic and treatment factors plus the number of biopsy cores collected and the number positive for malignancy by transrectal ultrasound-guided biopsy were available. The primary endpoint was biochemical failure (bF, Phoenix definition). Multivariate proportional hazards analyses were performed and expressed as a nomogram and the model's predictive ability assessed using the concordance index (c-index). Results: The cohort consisted of 21% low-, 51% intermediate-, and 28% high-risk cancer patients, and 30% had androgen deprivation with radiotherapy. The median PPC was 50% (interquartile range [IQR] 29-67%), and median follow-up was 51 months (IQR 29-71 months). Percentage of positive biopsy cores displayed an independent association with the risk of bF (p = 0.01), as did age, prostate-specific antigen value, Gleason score, clinical stage, androgen deprivation duration, and radiotherapy dose (p < 0.001 for all). Including PPC increased the c-index from 0.72 to 0.73 in the overall model. The influence of PPC varied significantly with radiotherapy dose and clinical stage (p = 0.02 for both interactions), with doses <66 Gy and palpable tumors showing the strongest relationship between PPC and bF. Intermediate-risk patients were poorly discriminated regardless of PPC inclusion (c-index 0.65 for both models). Conclusions: Outcome models incorporating PPC show only minor additional ability to predict biochemical failure beyond those containing standard prognostic factors.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Smart culture medium optimization for recombinant protein production: Experimental, modeling, and AI/ML-driven strategies

Recombinant protein production (RPP) is central to biotechnology, where recombinant proteins are used as either end products or catalysts in the synthesis of chemicals, fuels, and materials. Among the major cost drivers, culture medium plays a pivotal role in determining protein yield and quality. This review presents a comprehensive perspective on the critical stages of “smart” culture medium optimization: planning, screening, modeling, optimization, and validation. In the planning stage, we examine the nutritional and energetic roles of medium components, including carbon, nitrogen, amino acids, salts, and trace metals, and their impacts on culture parameters such as pH, oxidative state, and osmolality. We highlight the variability in trace metal content due to water sources, culture vessels, and raw materials, which can substantially influence RPP. The screening stage covers Design of Experiments (DoE) approaches, assessing their theoretical basis, implementation, and limitations. For modeling, we describe methods that integrate experimental data to develop predictive models for smart medium formulation. Model-based optimization strategies can then be employed to select optimal media compositions for a given application. The validation stage aims to evaluate model predictions and provide feedback for model training and refinement. Finally, we survey mechanistic and artificial intelligence/machine learning (AI/ML)-driven models as integrated, transformational tools for predictive modeling of bioprocess conditions, nutrient availability, cellular metabolism, and protein quality, with the goal of optimizing culture media to enhance protein yields while reducing costs and environmental impact. We conclude by addressing the challenges of translating laboratory-scale medium optimization to industrial-scale settings and exploring future AI/ML-driven approaches that may overcome current bottlenecks and accelerate medium design for RPP. Overall, this review provides a unified framework for advancing smart medium design in RPP.

Artificial Intelligence/Machine Learning (AI/ML)↗

Evaluating alternative ebullition models for predicting peatland methane emission and its pathways via data–model fusion

Abstract. Understanding the dynamics of peatland methane (CH4) emissions and quantifying sources of uncertainty in estimating peatland CH4 emissions are critical for mitigating climate change. The relative contributions of CH4 emission pathways through ebullition, plant-mediated transport, and diffusion, together with their different transport rates and vulnerability to oxidation, determine the quantity of CH4 to be oxidized before leaving the soil. Notwithstanding their importance, the relative contributions of the emission pathways are highly uncertain. In particular, the ebullition process is more uncertain and can lead to large uncertainties in modeled CH4 emissions. To improve model simulations of CH4 emission and its pathways, we evaluated two model structures: (1) the ebullition bubble growth volume threshold approach (EBG) and (2) the modified ebullition concentration threshold approach (ECT) using CH4 flux and concentration data collected in a peatland in northern Minnesota, USA. When model parameters were constrained using observed CH4 fluxes, the CH4 emissions simulated by the EBG approach (RMSE = 0.53) had a better agreement with observations than the ECT approach (RMSE = 0.61). Further, the EBG approach simulated a smaller contribution from ebullition but more frequent ebullition events than the ECT approach. The EBG approach yielded greatly improved simulations of pore water CH4 concentrations, especially in the deep soil layers, compared to the ECT approach. When constraining the EBG model with both CH4 flux and concentration data in model–data fusion, uncertainty of the modeled CH4 concentration profiles was reduced by 78 % to 86 % in comparison to constraints based on CH4 flux data alone. The improved model capability was attributed to the well-constrained parameters regulating the CH4 production and emission pathways. Our results suggest that the EBG modeling approach better characterizes CH4 emission and underlying mechanisms. Moreover, to achieve the best model results both CH4 flux and concentration data are required to constrain model parameterization.

59 BASIC BIOLOGICAL SCIENCES↗

REV-INR: Regularized Evidential Implicit Neural Representation for Uncertainty-Aware Volume Visualization

Applications of Implicit Neural Representations (INRs) have emerged as a promising deep learning approach for compactly representing large volumetric datasets. These models can act as surrogates for volume data, enabling efficient storage and on-demand reconstruction via model predictions. However, conventional deterministic INRs only provide value predictions without insights into the model’s prediction uncertainty or the impact of inherent noisiness in the data. This limitation can lead to unreliable data interpretation and visualization due to prediction inaccuracies in the reconstructed volume. Identifying erroneous results extracted from model-predicted data may be infeasible, as raw data may be unavailable due to its large size. To address this challenge, we introduce REV-INR, Regularized Evidential Implicit Neural Representation, which learns to predict data values accurately along with the associated coordinate-level data uncertainty and model uncertainty using only a single forward pass of the trained REV-INR during inference. By comprehensively comparing and contrasting REV-INR with existing well-established deep uncertainty estimation methods, we show that REV-INR achieves the best volume reconstruction quality with robust data (aleatoric) and model (epistemic) uncertainty estimates using the fastest inference time. Consequently, we demonstrate that REV-INR facilitates assessment of the reliability and trustworthiness of the extracted isosurfaces and volume visualization results, enabling analyses to be solely driven by model-predicted data.

Saklani, Shanu [Indian Institute of Technology, Ka↗