Search NASA⌕ Search

SEARCH · Search NASA

Results for “prior knowledge”

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 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Demonstration of machine-learning-enhanced Bayesian quantum state estimation

Machine learning (ML) has found broad applicability in quantum information science in topics as diverse as experimental design, state classification, and even studies on quantum foundations. Here, we experimentally realize an approach for defining custom prior distributions that are automatically tuned using ML for Bayesian quantum state estimation methods that generally better conform to the physical properties of the underlying system than standard fixed prior distributions. Previously, researchers have looked to Bayesian quantum state tomography for advantages like uncertainty quantification, the return of reliable estimates under any measurement condition, and minimal mean-squared error. However, practical challenges related to long computation times and conceptual issues concerning how to incorporate prior knowledge most suitably can overshadow these benefits. Using both simulated and experimental measurement results, we demonstrate that ML-defined prior distributions reduce net convergence times and provide a natural way to incorporate both implicit and explicit information directly into the prior distribution. These results constitute a promising path toward practical implementations of Bayesian quantum state tomography.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS1985

Experiments in structural dynamics and control using a grid

Future spacecraft are being conceived that are highly flexible and of extreme size. The two features of flexibility and size pose new problems in control system design. Since large scale structures are not testable in ground based facilities, the decision on component placement must be made prior to full-scale tests on the spacecraft. Control law research is directed at solving problems of inadequate modelling knowledge prior to operation required to achieve peak performance. Another crucial problem addressed is accommodating failures in systems with smart components that are physically distributed on highly flexible structures. Parameter adaptive control is a method of promise that provides on-orbit tuning of the control system to improve performance by upgrading the mathematical model of the spacecraft during operation. Two specific questions are answered in this work. They are: What limits does on-line parameter identification with realistic sensors and actuators place on the ultimate achievable performance of a system in the highly flexible environment? Also, how well must the mathematical model used in on-board analytic redundancy be known and what are the reasonable expectations for advanced redundancy management schemes in the highly flexible and distributed component environment?

Montgomery, R. C.↗
DOE OSTI2026

Deployment of Traditional and Hybrid Machine Learning for Critical Heat Flux Prediction in the CTF Thermal-Hydraulics Code

Critical heat flux (CHF) marks the transition from nucleate to film boiling, where heat transfer to the working fluid can rapidly deteriorate. Accurate CHF prediction is essential for efficiency, safety, and preventing equipment damage, particularly in nuclear reactors. Although widely used, empirical correlations frequently exhibit discrepancies when compared to experimental data, limiting their reliability in diverse operational conditions. Traditional machine learning (ML) approaches have demonstrated potential for CHF prediction but often suffer from limited interpretability, data scarcity, and insufficient knowledge of physical principles. Hybrid model approaches, which combine data-driven ML with base models, mitigate these concerns by incorporating prior knowledge of the domain. This study integrates an externally trained purely data-driven ML model and two hybrid models (using the Biasi and Bowring CHF correlations) within the CTF subchannel code via a custom Fortran framework. Performance was evaluated using two validation cases: a subset of the Nuclear Regulatory Commission (NRC) CHF database and the Bennett dryout experiments. In both cases, the hybrid models demonstrated significantly lower error metrics compared to conventional empirical correlations, with the best models often reducing relative error by about 5 percentage points. The pure ML model achieved comparable accuracy, outperforming the hybrid Biasi model in the NRC test case (3.3% versus 5.5% relative error) but exhibiting slightly higher error against the hybrid Bowring model in the Bennett test case (7.7% versus 6.1%). Trend analysis of error parity indicated that ML-based models reduced the tendency for CHF overprediction, improving overall accuracy. These results demonstrate that ML-based CHF models can be effectively integrated into subchannel codes and could potentially increase performance compared to conventional methods.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

Hierarchical Estimation For Planetary Protection

The software uses Bayes' theorem to describe the probability of an event based on prior knowledge of conditions that might be related to the event. The purpose of Bayesian analysis is to determine posterior probabilities based on prior probabilities where new information can be used in the decision-making process as additional data is gathered. The software will be used in Probabilistic Risk Assessments (PRAs) related to the Europa Clipper mission, which is one of NASA’s top priorities. Ultimately, the mission entails sending the Europa Clipper spacecraft to Jupiter’s Europa moon to orbit the planet and collect data for research and development. Europa is the smallest of the four Galilean moons orbiting Jupiter and is believed by researchers to be the most promising place to look for present-day environments suitable for life. Europa is thought to have an iron core, a rocky mantle, and a salt-water ocean covered by an ice-layered surface.

Gribok, Andrei [Idaho National Laboratory (INL), I↗
DOE OSTI2022

Physics makes the difference: Bayesian optimization and active learning via augmented Gaussian process

Abstract Both experimental and computational methods for the exploration of structure, functionality, and properties of materials often necessitate the search across broad parameter spaces to discover optimal experimental conditions and regions of interest in the image space or parameter space of computational models. The direct grid search of the parameter space tends to be extremely time-consuming, leading to the development of strategies balancing exploration of unknown parameter spaces and exploitation towards required performance metrics. However, classical Bayesian optimization (BO) strategies based on the Gaussian process (GP) do not readily allow for the incorporation of the known physical behaviors or past knowledge. Here we explore a hybrid optimization/exploration algorithm created by augmenting the standard GP with a structured probabilistic model of the expected system’s behavior. This approach balances the flexibility of the non-parametric GP approach with a rigid structure of physical knowledge encoded into the parametric model. The fully Bayesian treatment of the latter allows additional control over the optimization via the selection of priors for the model parameters. The method is demonstrated for a noisy version of a standard univariate test function used to evaluate optimization algorithms and further extended to physical lattice models. This methodology is expected to be universally suitable for injecting prior knowledge in the form of physical models and past data in the BO framework.

42 ENGINEERING↗
DOE OSTI2021

Predictability as a probe of manifest and latent physics: The case of atomic scale structural, chemical, and polarization behaviors in multiferroic Sm-doped BiFeO 3

The predictability of a certain effect or phenomenon is often equated with the knowledge of relevant physical laws, typically understood as a functional or numerically derived relationship between the observations and known states of the system. Correspondingly, observations inconsistent with prior knowledge can be used to derive new knowledge on the nature of the system or indicate the presence of yet unknown mechanisms. In this work, we explore the applicability of Gaussian processes (GP) to establish predictability and uncertainty of local behaviors from multimodal observations, providing an alternative to this classical paradigm. Using atomic resolution scanning transmission electron microscopy (STEM) of multiferroic Sm-doped BiFeO 3 across a broad composition range, we directly visualize the atomic structure and structural, physical, and chemical order parameter fields for the material. GP regression is used to establish the predictability of the local polarization field from different groups of parameters, including the adjacent polarization values and several combinations of physical and chemical descriptors, including lattice parameters, column intensities, etc. We observe that certain elements of microstructure, including charged and uncharged domain walls and interfaces with the substrate, are best predicted with specific combinations of descriptors, and this predictability and associated uncertainties are consistent across the composition series. The associated generative physical mechanisms are discussed. It is also found that certain parameter combinations tend to predict the orthorhombic phase in the cases where rhombohedral phase is observed, suggesting a potential role of clamping and confinement phenomena in phase equilibrium in Sm-BiFeO 3 system close to morphotropic phase boundary. We argue that predictability and uncertainty in observational data offer a new pathway to probe the physics of condensed matter systems from multimodal local observations.

74 ATOMIC AND MOLECULAR PHYSICS↗
NASA NTRS2012

An Elementary Overview of the Selection of Materials for Service in Oxygen-Enriched Environments

The process for selecting materials for use in oxygen or oxygen-enriched environments is one that continues to be investigated by many industries due to the importance to those industries of oxygen systems. There are several excellent resources available to assist oxygen systems design engineers and end-users, with the most comprehensive being ASTM MNL-36, Safe Use of Oxygen and Oxygen Systems: Handbook for Design, Operation and Maintenance, 2nd Edition. ASTM also makes available several standards for oxygen systems. However, the ASTM publications are extremely detailed, and typically designed for professionals who already possess a working knowledge of oxygen systems. No notable resource exists, whether an ASTM or other organizational publication, which can be used to educate engineers or technicians who have no prior knowledge of the nuances of oxygen system design and safety. This paper will fill the void for information needed by organizations that design or operate oxygen systems. The information in this paper is not new information, but is a concise and easily understood summary of selecting materials for oxygen systems. This paper will serve well as an employee s first introduction to oxygen system materials selection, and probably the employee s first introduction to ASTM.

Davis, Samuel Eddie↗
NASA NTRS1994

A User-Centered Approach to Adaptive Hypertext Based on an Information Relevance Model

Rapid and effective to information in large electronic documentation systems can be facilitated if information relevant in an individual user's content can be automatically supplied to this user. However most of this knowledge on contextual relevance is not found within the contents of documents, it is rather established incrementally by users during information access. We propose a new model for interactively learning contextual relevance during information retrieval, and incrementally adapting retrieved information to individual user profiles. The model, called a relevance network, records the relevance of references based on user feedback for specific queries and user profiles. It also generalizes such knowledge to later derive relevant references for similar queries and profiles. The relevance network lets users filter information by context of relevance. Compared to other approaches, it does not require any prior knowledge nor training. More importantly, our approach to adaptivity is user-centered. It facilitates acceptance and understanding by users by giving them shared control over the adaptation without disturbing their primary task. Users easily control when to adapt and when to use the adapted system. Lastly, the model is independent of the particular application used to access information, and supports sharing of adaptations among users.

Mathe, Nathalie↗
NASA NTRS1975

Graphical derivations of radar, sonar, and communication signals

The designer of a communication system often has knowledge concerning the changes in distance between transmitter and receiver as a function of time. This information can be exploited to reduce multipath interference via proper signal design. A radar or sonar may also have good a priori information about possible target trajectories. Such knowledge can again be used to reduce the receiver's response to clutter (MTI), to enhance signal-to-noise ratio, or to simplify receiver design. There are also situations in which prior knowledge about trajectories is lacking. The system should then utilize a single-filter pair which is insensitive to the effects induced by relative motion between transmitter, receiver, and reflectors. For waveforms with large time-bandwidth products, such as long pulse trains, it is possible to graphically derive signal formats for both situations (trajectory known and unknown). Although the exact form of the signal is sometimes not specified by the graphical procedure, the problem in such cases is reduced to one which has already been solved, i.e., the generation of an impulse equivalent code.

Altes, R. A.↗
NASA NTRS2002

Incorporating Biological Knowledge into Evaluation of Casual Regulatory Hypothesis

Biological data can be scarce and costly to obtain. The small number of samples available typically limits statistical power and makes reliable inference of causal relations extremely difficult. However, we argue that statistical power can be increased substantially by incorporating prior knowledge and data from diverse sources. We present a Bayesian framework that combines information from different sources and we show empirically that this lets one make correct causal inferences with small sample sizes that otherwise would be impossible.

Chrisman, Lonnie↗
DOE OSTI2024

The Effect of the Prior and the Experimental Design on the Inference of the Precision Matrix in Gaussian Chain Graph Models

Here, we investigate whether (and how) experimental design could aid in the estimation of the precision matrix in a Gaussian chain graph model, especially the interplay between the design, the effect of the experiment and prior knowledge about the effect. Estimation of the precision matrix is a fundamental task to infer biological graphical structures like microbial networks. We compare the marginal posterior precision of the precision matrix under four priors: flat, conjugate Normal-Wishart, Normal-MGIG and a general independent. Under the flat and conjugate priors, the Laplace-approximated posterior precision is not a function of the design matrix rendering useless any efforts to find an optimal experimental design to infer the precision matrix. In contrast, the Normal-MGIG and general independent priors do allow for the search of optimal experimental designs, yet there is a sharp upper bound on the information that can be extracted from a given experiment. We confirm our theoretical findings via a simulation study comparing (i) the KL divergence between prior and posterior and (ii) the Stein’s loss difference of MAPs between random and no experiment. Our findings provide practical advice for domain scientists conducting experiments to better infer the precision matrix as a representation of a biological network.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Evaluating disease surveillance strategies for early outbreak detection in contact networks with varying community structure

Disease surveillance systems allow public health agencies to respond to emerging diseases before they become widespread. Developing such systems requires identifying optimal ways to monitor in the context of an epidemic outbreak; this problem is known as sensor selection. Contact networks represent the dynamics of interaction in a population and are used to model how a disease spreads in a population and to explore strategies of sensor selection. We evaluated five sensor selection strategies on their ability to provide an early warning of a COVID-like outbreak in synthetic contact networks encapsulated in four network scenarios. Three of these scenarios assessed different aspects of community structure. The fourth scenario employed a contact network representing the population and interactions of 6.8 million people in New York City, constructed from an agent-based simulation using census and transportation data. This scenario exemplifies how sensor selection strategies may perform in a real-world, urban context. Our findings suggest that the choice of the optimal strategy depends heavily on the community structure of the network. Strategies that select highly connected nodes or maximize network coverage are the optimal surveillance strategy for outbreak detection in many network community structures. However, a naive implementation of these strategies may fail to provide an early warning at all—including in the New York City scenario. Moreover, these methods are impractical for real-world use as they require knowledge of the underlying contact network. Instead, a selection strategy that starts with a set of random nodes and then performs a random walk through a chain of neighbors reliably provides early warnings without requiring prior knowledge of the network. We find this method, called “random chain”, to be the most pragmatic for implementation in a real-world disease surveillance context.

60 APPLIED LIFE SCIENCES↗
NASA NTRS2015

Flexible Micropost Arrays for Shear Stress Measurement

Increased fuel costs, heightened environmental protection requirements, and noise abatement continue to place drag reduction at the forefront of aerospace research priorities. Unfortunately, shortfalls still exist in the fundamental understanding of boundary-layer airflow over aerodynamic surfaces, especially regarding drag arising from skin friction. For example, there is insufficient availability of instrumentation to adequately characterize complex flows with strong pressure gradients, heat transfer, wall mass flux, three-dimensionality, separation, shock waves, and transient phenomena. One example is the acoustic liner efficacy on aircraft engine nacelle walls. Active measurement of shear stress in boundary layer airflow would enable a better understanding of how aircraft structure and flight dynamics affect skin friction. Current shear stress measurement techniques suffer from reliability, complexity, and airflow disruption, thereby compromising resultant shear stress data. The state-of-the-art for shear stress sensing uses indirect or direct measurement techniques. Indirect measurements (e.g., hot-wire, heat flux gages, oil interferometry, laser Doppler anemometry, small scale pressure drag surfaces, i.e., fences) require intricate knowledge of the studied flow, restrictive instrument arrangements, large surface areas, flow disruption, or seeding material; with smaller, higher bandwidth probes under development. Direct measurements involve strain displacement of a sensor element and require no prior knowledge of the flow. Unfortunately, conventional "floating" recessed components for direct measurements are mm to cm in size. Whispering gallery mode devices and Fiber Bragg Gratings are examples of recent additions to this type of sensor with much smaller (m) sensor components. Direct detection techniques are often single point measurements and difficult to calibrate and implement in wind tunnel experiments. In addition, the wiring, packaging, and installation of delicate micro-electromechanical devices impede the use of most direct shear sensors. Similarly, the cavity required for sensing element displacement is sensitive to particulate obstruction. This work was focused on developing a shear stress sensor for use in subsonic wind tunnel test facilities applicable to an array of test configurations. The non-displacement shear sensors described here have minimal packaging requirements resulting in minimal or no disturbance of boundary layer flow. Compared to previous concepts, device installation could be simple with reduced cost and down-time. The novelty lies in the creation of low profile (nanoscale to 100 μm) micropost arrays that stay within the viscous sub-layer of the airflow. Aerodynamic forces, which are related to the surface shear stress, cause post deflection and optical property changes. Ultimately, a reliable, accurate shear stress sensor that does not disrupt the airflow has the potential to provide high value data for flow physics researchers, aerodynamicists, and aircraft manufacturers leading to greater flight efficiency arising from more in-depth knowledge on how aircraft design impacts near surface properties.

Wohl, Christopher J.↗
DOE OSTI2021

Autonomous extraction of millimeter-scale deformation in InSAR time series using deep learning

Systematically characterizing slip behaviours on active faults is key to unraveling the physics of tectonic faulting and the interplay between slow and fast earthquakes. Interferometric Synthetic Aperture Radar (InSAR), by enabling measurement of ground deformation at a global scale every few days, may hold the key to those interactions. However, atmospheric propagation delays often exceed ground deformation of interest despite state-of-the art processing, and thus InSAR analysis requires expert interpretation and a priori knowledge of fault systems, precluding global investigations of deformation dynamics. Here, we show that a deep auto-encoder architecture tailored to untangle ground deformation from noise in InSAR time series autonomously extracts deformation signals, without prior knowledge of a fault’s location or slip behaviour. Applied to InSAR data over the North Anatolian Fault, our method reaches 2 mm detection, revealing a slow earthquake twice as extensive as previously recognized. We further explore the generalization of our approach to inflation/deflation-induced deformation, applying the same methodology to the geothermal field of Coso, California.

58 GEOSCIENCES↗
NASA NTRS2015

Pre-Mission Input Requirements to Enable Successful Sample Collection by a Remote Field/EVA Team

This paper is intended to evaluate the sample collection process with respect to sample characterization and decision making. In some cases, it may be sufficient to know whether a given outcrop or hand sample is the same as or different from previous sampling localities or samples. In other cases, it may be important to have more in-depth characterization of the sample, such as basic composition, mineralogy, and petrology, in order to effectively identify the best sample. Contextual field observations, in situ/handheld analysis, and backroom evaluation may all play a role in understanding field lithologies and their importance for return. For example, whether a rock is a breccia or a clast-laden impact melt may be difficult based on a single sample, but becomes clear as exploration of a field site puts it into context. The FINESSE (Field Investigations to Enable Solar System Science and Exploration) team is a new activity focused on a science and exploration field based research program aimed at generating strategic knowledge in preparation for the human and robotic exploration of the Moon, near-Earth asteroids (NEAs) and Phobos and Deimos. We used the FINESSE field excursion to the West Clearwater Lake Impact structure (WCIS) as an opportunity to test factors related to sampling decisions. In contract to other technology-driven NASA analog studies, The FINESSE WCIS activity is science-focused, and moreover, is sampling-focused, with the explicit intent to return the best samples for geochronology studies in the laboratory. This specific objective effectively reduces the number of variables in the goals of the field test and enables a more controlled investigation of the role of the crewmember in selecting samples. We formulated one hypothesis to test: that providing details regarding the analytical fate of the samples (e.g. geochronology, XRF/XRD, etc.) to the crew prior to their traverse will result in samples that are more likely to meet specific analytical objectives than samples collected in the absence of this premission information. We conducted three tests of this hypothesis. Our investigation was designed to document processes, tools and procedures for crew sampling of planetary targets. This is not meant to be a blind, controlled test of crew efficacy, but rather an effort to recognize the relevant variables that enter into sampling protocol and to develop recommendations for crew and backroom training in future endeavors. Methods: One of the primary FINESSE field deployment objectives was to collect impact melt rocks and impact melt-bearing breccias from a number of locations around the WCIS structure to enable high precision geochronology of the crater to be performed [1]. We conducted three tests at WCIS after two full days of team participation in field site activities, including using remote sensing data and geologic maps, hiking overland to become familiar with the terrain, and examining previously-collected samples from other islands. In addition, the team members shared their projects and techniques with the entire team. We chose our "crew members" as volunteers from the team, all of whom had had moderate training in geologic fieldwork and became familiar with the general field setting. The first two tests were short, focused tests of our hypothesis. Test A was to obtain hydrothermal vugs; Test B was to obtain impact melt and intrusive rock as well as the contact between the two to check for contact metamorphism and age differences. In both cases, the test director had prior knowledge of the site geology and had developed a study-specific objective for sampling prior to deployment. Prior to the field deployment, the crewmember was briefed on the sampling objective and the laboratory techniques that would be used on the samples. At the field sites (Fig. 2), the crewmember was given 30 minutes to survey a small section of outcrop (10-15 m) and acquire a suite of three samples. The crewmember talked through his process and the test director kept track of the timeline in verbal cues to the crewmember. At the conclusion, the team member conducting the scientific study appraised the samples and train of thought. Test C was a 90-minute EVA simulation using two crewmembers working out of line-of-sight in communication with a science backroom. The science objectives were determined by the science backroom team in advance using a Gigapan image of the outcrop (Fig. 1). The science team formulated hypotheses for the outcrop units and created sampling objectives for impact-melt lithologies; the science team turned these into a science plan, which they communicated to the crew in camp prior to crew deployment. As part of the science plan, the science team also discussed their sample needs in depth with the crewmembers, including laboratory methods, objectives, and samples sizes needed. During the deployment, the two crewmembers relayed real-time information to the science backroom by radio with no time delay. Both the crew and science team re-evaluated their hypotheses and science plans in real-time. Discussion: Upon evaluation, we found that the focused tests (Tests A and B) were successful in meeting their scientific objectives. The crewmember used their knowledge of how the samples were to be used in further study (technique, sample size, and scientific need) to focus on the sampling task. The crewmember was comfortable spending minimal time describing and mapping the outcrop. The crewmember used all available time to get a good sample. The larger test was unsuccessful in meeting the sampling objectives. When the crewmembers began describing the lithologies, it was quickly apparent that the lithologies were not as the backroom expected and had communicated to the crew. When the outcrop wasn't as expected, the crew members instinctively switched to field characterization mode, taking significant time to characterize and map the outcrop. One crew member admitted that he "kind of lost track" of the sampling strategy as he focused on the basic outcrop characterization. This is the logical first step in a field geology campaign, that a significant amount of time must be spent by the crew and backroom to understand the outcrop and its significance. Basic field characterization of an outcrop is a focused activity that takes significant time and training [2, 3]. Sampling of representational lithologies can be added to this activity for little cost [4]. However, we have shown that identification of unusual or specific samples for laboratory study also takes significant time and knowledge. We suggest that sampling of this type be considered a separate activity from field characterization, and that crewmembers be trained in sampling needs for different kinds of studies (representative lithologies vs. specialized samples) to acquire a mindset for sampling similar to field mapping. Sampling activities should be given a significant amount of specifically allocated time in scheduling EVA activities; and in the better case, that sampling be done as a second activity to a previously studied outcrop where both crew and backroom are comfortable with its context and characteristics. Our hypothesis posited that crewmember knowledge of how the samples would be used upon return would aid them in choosing relevant samples. Our testing bore this hypothesis out to some extent. We therefore recommend that crewmember training should include exposure to the laboratory techniques and analyses that will be used on the samples to foster this knowledge. There is also the potential for increasing crewmember contextual knowledge real-time in the field through the introduction of in situ geochemical technologies such as field portable XRF. The presence of field portable geochemical technology could enable the astronauts to interrogate the samples for K abundance real-time, ensuring they could collect valuable and dateable samples [5]. Though simulations such as these can teach us a fair bit about decision making processes and timeline building, one EVA participant noted that when he wasn't collecting "real" samples, he wasn't at his best. This effect suggests that higher-fidelity studies involving truly remote participants conducting actual scientific studies merit further attention to capture lessons for application to future crew situations.

sample collection↗
DOE OSTI2024

Information-incorporated gene network construction with FDR control

Abstract Motivation Large-scale gene expression studies allow gene network construction to uncover associations among genes. To study direct associations among genes, partial correlation-based networks are preferred over marginal correlations. However, FDR control for partial correlation-based network construction is not well-studied. In addition, currently available partial correlation-based methods cannot take existing biological knowledge to help network construction while controlling FDR. Results In this paper, we propose a method called Partial Correlation Graph with Information Incorporation (PCGII). PCGII estimates partial correlations between each pair of genes by regularized node-wise regression that can incorporate prior knowledge while controlling the effects of all other genes. It handles high-dimensional data where the number of genes can be much larger than the sample size and controls FDR at the same time. We compare PCGII with several existing approaches through extensive simulation studies and demonstrate that PCGII has better FDR control and higher power. We apply PCGII to a plant gene expression dataset where it recovers confirmed regulatory relationships and a hub node, as well as several direct associations that shed light on potential functional relationships in the system. We also introduce a method to supplement observed data with a pseudogene to apply PCGII when no prior information is available, which also allows checking FDR control and power for real data analysis. Availability and implementation R package is freely available for download at https://cran.r-project.org/package=PCGII.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Preparing for the next pandemic via transfer learning from existing diseases with hierarchical multi-modal BERT: a study on COVID-19 outcome prediction

Abstract Developing prediction models for emerging infectious diseases from relatively small numbers of cases is a critical need for improving pandemic preparedness. Using COVID-19 as an exemplar, we propose a transfer learning methodology for developing predictive models from multi-modal electronic healthcare records by leveraging information from more prevalent diseases with shared clinical characteristics. Our novel hierarchical, multi-modal model ( $${\textsc {TransMED}}$$ T R A N S MED ) integrates baseline risk factors from the natural language processing of clinical notes at admission, time-series measurements of biomarkers obtained from laboratory tests, and discrete diagnostic, procedure and drug codes. We demonstrate the alignment of $${\textsc {TransMED}}$$ T R A N S MED ’s predictions with well-established clinical knowledge about COVID-19 through univariate and multivariate risk factor driven sub-cohort analysis. $${\textsc {TransMED}}$$ T R A N S MED ’s superior performance over state-of-the-art methods shows that leveraging patient data across modalities and transferring prior knowledge from similar disorders is critical for accurate prediction of patient outcomes, and this approach may serve as an important tool in the early response to future pandemics.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Decoding the protein–ligand interactions using parallel graph neural networks

Abstract Protein–ligand interactions (PLIs) are essential for biochemical functionality and their identification is crucial for estimating biophysical properties for rational therapeutic design. Currently, experimental characterization of these properties is the most accurate method, however, this is very time-consuming and labor-intensive. A number of computational methods have been developed in this context but most of the existing PLI prediction heavily depends on 2D protein sequence data. Here, we present a novel parallel graph neural network (GNN) to integrate knowledge representation and reasoning for PLI prediction to perform deep learning guided by expert knowledge and informed by 3D structural data. We develop two distinct GNN architectures: $$\hbox {GNN}_{\mathrm{F}}$$ GNN F is the base implementation that employs distinct featurization to enhance domain-awareness, while $$\hbox {GNN}_{\mathrm{P}}$$ GNN P is a novel implementation that can predict with no prior knowledge of the intermolecular interactions. The comprehensive evaluation demonstrated that GNN can successfully capture the binary interactions between ligand and protein’s 3D structure with 0.979 test accuracy for $$\hbox {GNN}_{\mathrm{F}}$$ GNN F and 0.958 for $$\hbox {GNN}_{\mathrm{P}}$$ GNN P for predicting activity of a protein–ligand complex. These models are further adapted for regression tasks to predict experimental binding affinities and $$\hbox {pIC}_{\mathrm{50}}$$ pIC 50 crucial for compound’s potency and efficacy. We achieve a Pearson correlation coefficient of 0.66 and 0.65 on experimental affinity and 0.50 and 0.51 on $$\hbox {pIC}_{\mathrm{50}}$$ pIC 50 with $$\hbox {GNN}_{\mathrm{F}}$$ GNN F and $$\hbox {GNN}_{\mathrm{P}}$$ GNN P , respectively, outperforming similar 2D sequence based models. Our method can serve as an interpretable and explainable artificial intelligence (AI) tool for predicted activity, potency, and biophysical properties of lead candidates. To this end, we show the utility of $$\hbox {GNN}_{\mathrm{P}}$$ GNN P on SARS-Cov-2 protein targets by screening a large compound library and comparing the prediction with the experimentally measured data.

59 BASIC BIOLOGICAL SCIENCES↗