Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model selection”

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 667 records · Page 37
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Towards Anomaly Detection at the CMS High-Level Trigger System

Traditional trigger strategies in CMS typically rely on model-dependent selections or rigid kinematic cuts, risking the omission of unexpected exotic signatures. To address this, we propose a novel anomaly detection (AD) algorithm for the High-Level Trigger (HLT), designed to serve as a complementary second layer of filtering to the Level-1 AXOL1TL AD algorithm. We employ a transformer-based foundation model trained on a diverse ensemble of Standard Model processes. By combining a joint contrastive and classification objective, and using particle kinematics as inputs, the model learns to map events to a physics-informed latent space where anomalous events are isolated from dominant backgrounds. Preliminary results show that this strategy enhances the signal-to-background ratio across a range of rare SM and BSM scenarios. Furthermore, this work constitutes foundational R&D for the potential implementation of an analogous AD algorithm in the Level-1 trigger system for Phase-2.

Cruz, Roy [U. Wisconsin, Madison (main)] (ORCID:00↗
DOE OSTI2020

Adapting Existing Energy Planning, Simulation, and Operational Models for Resilience Analysis

Resilience is a topic receiving much attention in relation to energy systems, with particular attention being paid to the supply of electricity. As a result, research communities have proposed a plethora of candidate indicators and metrics for resilience, most of which remain immature at different scales and segments within the energy system. Given the complexity of resilience analyses and mitigation strategies, there is limited value in attempting to identify a single resilience metric, as no one metric can quantify resilience or its associated value for all stakeholders. Instead, a necessary focus of the research community should lie in implementing, testing, and validating resilience metrics and analysis approaches in energy sector models, which will be invaluable for informing resilience planning and investment decisions. Recognizing that implementing, testing, and validating resilience metrics are challenges that need to be addressed, the National Renewable Energy Laboratory (NREL) dedicated staff and time to researching how to effectively integrate resilience considerations into energy sector models and tools, as part of the Laboratory Directed Research and Development (LDRD) program. The overarching goal of the effort was to evaluate the data needs, methodologies, and outcomes - including consequences and/or changes in investment or operational decisions due to avoided consequences - based on resilience analysis in a range of existing tools. In particular, we selected five models originally built at NREL to explore non-resilience energy research questions to implement and exercise resilience metrics and analysis approaches.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
NASA NTRS1976

Aeroheating model advancements featuring electroless metallic plating

Discussed are advancements in wind tunnel model construction methods and hypersonic test data demonstrating the methods. The general objective was to develop model fabrication methods for improved heat transfer measuring capability at less model cost. A plated slab model approach was evaluated with cast models containing constantan wires that formed single-wire-to-plate surface thermocouple junctions with a seamless skin of electroless nickel alloy. The surface of a space shuttle orbiter model was selectively plated with scaled tiles to simulate, with high fidelity, the probable misalignments of the heatshield tiles on a flight vehicle. Initial, Mach 8 heating results indicated a minor effect of tile misalignment roughness on boundary layer transition, implying a possible relaxation of heatshield manufacturing tolerances. Some loss of the plated tiles was experienced when the model was tested at high heating rates.

Stalmach, C. J., Jr.↗
NASA NTRS1987

Life prediction and constitutive models for engine hot section anisotropic materials

The results are presented of a program designed to develop life prediction and constitutive models for two coated single crystal alloys used in gas turbine airfoils. The two alloys are PWA 1480 and Alloy 185. The two oxidation resistant coatings are PWA 273, an aluminide coating, and PWA 286, an overlay NiCoCrAlY coating. To obtain constitutive and fatigue data, tests were conducted on uncoated and coated specimens loaded in the CH76 100 CH110 , CH76 110 CH110 , CH76 111 CH110 and CH76 123 CH110 crystallographic directions. Two constitutive models are being developed and evaluated for the single crystal materials: a micromechanic model based on crystallographic slip systems, and a macroscopic model which employs anisotropic tensors to model inelastic deformation anisotropy. Based on tests conducted on the overlay coating material, constitutive models for coatings also appear feasible and two initial models were selected. A life prediction approach was proposed for coated single crystal materials, including crack initiation either in the coating or in the substrate. The coating initiated failures dominated in the tests at load levels typical of gas turbine operation. Coating life was related to coating stress/strain history which was determined from specimen data using the constitutive models.

Swanson, G. A.↗
NASA NTRS1995

Mathematical Model Of Curing Behavior Of A Polymer

Mathematical model predicts selected aspects of chemical, thermal and mechanical responses of polymeric liner and propellant materials during curing process. Predictions made both prior to processing and during process in quasi-real time. Developed specifically for use in designing and analyzing process in which bondline materials (including polymeric liners) and propellant cast and cured in rocket motor. With modifications, model applicable to curing of other polymeric materials. Further development may provide direct "in-line" program for calculations, including comparisons, in real time. This will constitute basis for more sophisticated control of process.

Loomis, Willard C.↗
NASA NTRS2022

SKYSURF: Constraints on Zodiacal Light and Extragalactic Background Light through Panchromatic HST All-sky Surface-brightness Measurements: II. First Limits on Diffuse Light at 1.25, 1.4, and 1.6 μm

We present the first results from the HST Archival Legacy project "SKYSURF." As described in Windhorst et al., SKYSURF utilizes the large HST archive to study the diffuse UV, optical, and near-IR backgrounds and foregrounds in detail. Here, we utilize SKYSURF's first sky-surface-brightness measurements to constrain the level of near-IR diffuse Extragalactic Background Light (EBL) in three near-IR filters (F125W, F140W, and F160W). This is done by comparing our preliminary sky measurements of >30,000 images to zodiacal light models, carefully selecting the darkest images to avoid contamination from stray light. Our sky-surface-brightness measurements have been verified to an accuracy of better than 1%, which when combined with systematic errors associated with HST, results in sky-brightness uncertainties of ∼ 2%–4% ≃ 0.005 MJy sr−1 in each image. When compared to the Kelsall et al. zodiacal model, an isotropic diffuse background of ∼30 nW m−2 sr−1 remains, whereas using the Wright zodiacal model results in no discernible diffuse background. Based primarily on uncertainties in the foreground model subtraction, we present limits on the amount of diffuse EBL of 29, 40, and 29 nW m−2 sr−1, for F125W, F140W, and F160W, respectively. While this light is generally isotropic, our modeling at this point does not distinguish between a cosmological origin or a solar system origin (such as a dim, diffuse, spherical cloud of cometary dust).

Timothy Carleton↗
DOE OSTI2024

Training custom light curve models of SN Ia subpopulations selected according to host galaxy properties

ABSTRACT Type Ia supernova (SN Ia) cosmology analyses include a luminosity step function in their distance standardization process to account for an observed yet unexplained difference in the post-standardization luminosities of SNe Ia originating from different host galaxy populations [e.g. high-mass ($M \gtrsim 10^{10} \, {\rm M}_{\odot }$) versus low-mass galaxies]. We present a novel method for including host-mass correlations in the SALT3 (Spectral Adaptive Light curve Template 3) light curve model used for standardizing SN Ia distances. We split the SALT3 training sample according to host-mass, training independent models for the low- and high-host-mass samples. Our models indicate that there are different average Si ii spectral feature strengths between the two populations, and that the average spectral energy distribution of SNe from low-mass galaxies is bluer than the high-mass counterpart. We then use our trained models to perform an SN cosmology analysis on the 3-yr spectroscopically confirmed Dark Energy Survey SN sample, treating SNe from low- and high-mass host galaxies as separate populations throughout. We find that our mass-split models reduce the Hubble residual scatter in the sample, albeit at a low statistical significance. We do find a reduction in the mass-correlated luminosity step but conclude that this arises from the model-dependent re-definition of the fiducial SN absolute magnitude rather than the models themselves. Our results stress the importance of adopting a standard definition of the SN parameters (x0, x1, c) in order to extract the most value out of the light curve modelling tools that are currently available and to correctly interpret results that are fit with different models.

Taylor, G. (ORCID:0000000157563259)↗
DOE OSTI2024

Emulator-Based Bayesian Calibration of the CISNET Colorectal Cancer Models

Purpose To calibrate Cancer Intervention and Surveillance Modeling Network (CISNET)'s SimCRC, MISCAN-Colon, and CRC-SPIN simulation models of the natural history colorectal cancer (CRC) with an emulator-based Bayesian algorithm and internally validate the model-predicted outcomes to calibration targets.Methods We used Latin hypercube sampling to sample up to 50,000 parameter sets for each CISNET-CRC model and generated the corresponding outputs. We trained multilayer perceptron artificial neural networks (ANNs) as emulators using the input and output samples for each CISNET-CRC model. We selected ANN structures with corresponding hyperparameters (i.e., number of hidden layers, nodes, activation functions, epochs, and optimizer) that minimize the predicted mean square error on the validation sample. We implemented the ANN emulators in a probabilistic programming language and calibrated the input parameters with Hamiltonian Monte Carlo-based algorithms to obtain the joint posterior distributions of the CISNET-CRC models' parameters. We internally validated each calibrated emulator by comparing the model-predicted posterior outputs against the calibration targets.Results The optimal ANN for SimCRC had 4 hidden layers and 360 hidden nodes, MISCAN-Colon had 4 hidden layers and 114 hidden nodes, and CRC-SPIN had 1 hidden layer and 140 hidden nodes. The total time for training and calibrating the emulators was 7.3, 4.0, and 0.66 h for SimCRC, MISCAN-Colon, and CRC-SPIN, respectively. The mean of the model-predicted outputs fell within the 95% confidence intervals of the calibration targets in 98 of 110 for SimCRC, 65 of 93 for MISCAN, and 31 of 41 targets for CRC-SPIN.Conclusions Using ANN emulators is a practical solution to reduce the computational burden and complexity for Bayesian calibration of individual-level simulation models used for policy analysis, such as the CISNET CRC models. In this work, we present a step-by-step guide to constructing emulators for calibrating 3 realistic CRC individual-level models using a Bayesian approach.

artificial neural networks↗
DOE OSTI2026

Towards Spatial Control of Reaction Selectivity on Photocatalysts using Area Selective Atomic Layer Deposition on Model Dual Site Electrocatalyst Platform

Photocatalytic water splitting is a promising route to low-cost H2; however, this approach is currently limited by solar-to-hydrogen (STH) conversion efficiencies in the sub-10% range. Z-scheme water splitting, in which H2 and O2 evolving particles are operated in separate compartments and electronically coupled by a soluble redox mediator, offers improvements in STH efficiency but introduces high rates of undesired side and back reactions. Nanoscopic oxide coatings (e.g., CrOx, and SiOx) have previously been used to selectively block undesired reactants from reaching active sites; however, a coating encapsulating the entire photocatalyst particle limits activity as the particle can no longer facilitate both half reactions. In contrast to photodeposition, which may produce non-uniform overlayers due to the difficulty of controlling local electrochemical reactivity on the surface, area-selective atomic layer deposition (AS-ALD) can be used to deposit conformal ultrathin oxide coatings while preserving access to non-growth areas of a substrate surface. To develop this technique for Z-scheme photocatalysts, we performed AS-ALD of TiO2 on a dual site planar electrocatalyst based on interdigitated arrays of Pt and Au. Self-assembled monolayers of 1-octadecanethiol (ODT) were used to block growth on the Au array, resulting in a patterned surface in which only the Pt sites were encapsulated with TiO2. These electrocatalysts demonstrated localized reaction selectivity for hydrogen evolution reaction (HER) over the TiO2/Pt sites, while the uncoated Au sites retained activity towards Fe(II)/Fe(III) redox (FeRR). Under independent potential control, these microelectrodes showed that selectively deposited TiO2 coatings can suppress the rate of back reactions on neighboring active sites by an order of magnitude compared to uncoated control samples.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Uncertainty-aware mixed-variable machine learning for materials design

Abstract Data-driven design shows the promise of accelerating materials discovery but is challenging due to the prohibitive cost of searching the vast design space of chemistry, structure, and synthesis methods. Bayesian optimization (BO) employs uncertainty-aware machine learning models to select promising designs to evaluate, hence reducing the cost. However, BO with mixed numerical and categorical variables, which is of particular interest in materials design, has not been well studied. In this work, we survey frequentist and Bayesian approaches to uncertainty quantification of machine learning with mixed variables. We then conduct a systematic comparative study of their performances in BO using a popular representative model from each group, the random forest-based Lolo model (frequentist) and the latent variable Gaussian process model (Bayesian). We examine the efficacy of the two models in the optimization of mathematical functions, as well as properties of structural and functional materials, where we observe performance differences as related to problem dimensionality and complexity. By investigating the machine learning models’ predictive and uncertainty estimation capabilities, we provide interpretations of the observed performance differences. Our results provide practical guidance on choosing between frequentist and Bayesian uncertainty-aware machine learning models for mixed-variable BO in materials design.

36 MATERIALS SCIENCE↗
NASA NTRS1984

Heat and mass transfer in combustion - Fundamental concepts and analytical techniques

Fundamental combustion phenomena and the associated flame structures in laminar gaseous flows are discussed on physical bases within the framework of the three nondimensional parameters of interest to heat and mass transfer in chemically-reacting flows, namely the Damkoehler number, the Lewis number, and the Arrhenius number which is the ratio of the reaction activation energy to the characteristic thermal energy. The model problems selected for illustration are droplet combustion, boundary layer combustion, and the propagation, flammability, and stability of premixed flames. Fundamental concepts discussed include the flame structures for large activation energy reactions, S-curve interpretation of the ignition and extinctin states, reaction-induced local-similarity and non-similarity in boundary layer flows, the origin and removal of the cold boundary difficulty in modeling flame propagation, and effects of flame stretch and preferential diffusion on flame extinction and stability. Analytical techniques introduced include the Shvab-Zeldovich formulation, the local Shvab-Zeldovich formulation, flame-sheet approximation and the associated jump formulation, and large activation energy matched asymptotic analysis. Potentially promising research areas are suggested.

Law, C. K.↗
NASA NTRSDate not supplied

Analysis of Oxygen Liquefaction with Transient Flow Rates for ISRU Systems

Human occupation of lunar and Martian surfaces requires in-situ resource utilization (ISRU) to create a sustainable environment with the limited resources available in space. Fuel and oxidizer generation is essential for developing a refueling capability for spacecraft, significantly reducing the propellant mass required for landing. Liquefaction systems are a key step in producing and storing cryogenic liquids such as oxygen, hydrogen, and methane. All those fluids are critical to propulsion, life support, and other spacecraft systems. NASA has previously confirmed the presence of water on the Moon and an electrolysis process can be used to separate the oxygen from the hydrogen molecules. Oxygen has also been found within lunar regolith and can be separated through ISRU processes. The extracted oxygen gas can then be liquefied inside a storage tank with tube-on-tank heat exchangers for future use as an oxidizer for propellants. While steady state liquefaction performance is easy to model, power limitations and cyclical environments caused by the change in Sun exposure between day and night periods potentially call for transient operations. The following analysis was performed using the Thermal Desktop software to investigate how various transient gaseous oxygen (GOX) flow rates impact the production and storage of liquid oxygen (LOX) in a 1-g liquefaction system. Variations of sinusoidal and exponential step functions were selected to model potential GOX flow rates that could be experienced by ISRU systems on lunar or Martian surfaces. This early computational analysis provides some insight on how transient operations impact oxygen liquefaction systems and helps illustrate operational questions to explore experimentally throughout the design process.

Cryogenic Fluid Management↗
DOE OSTI2025

Guiding Principles for Geochemical/Thermodynamic Model Development and Validation in Nuclear Waste Disposal: A Close Examination of Recent Thermodynamic Models for H + —Nd 3+ —NO 3 - (—Oxalate) Systems

Development of a defensible source-term model (STM), usually a thermodynamical model for radionuclide solubility calculations, is critical to a performance assessment (PA) of a geologic repository for nuclear waste disposal. Such a model is generally subjected to rigorous regulatory scrutiny. In this article, we highlight key guiding principles for STM model development and validation in nuclear waste management. We illustrate these principles by closely examining three recently developed thermodynamic models with the Pitzer formulism for aqueous H + —Nd 3+ —NO 3 - (—oxalate) systems in a reverse alphabetical order of the authors: the XW model developed by Xiong and Wang, the OWC model developed by Oakes et al., and the GLC model developed by Guignot et al., among which the XW model deals with trace activity coefficients for Nd(III), while the OWC and GLC models are for concentrated Nd(NO 3 ) 3 electrolyte solutions. The principles highlighted include the following: (1) Principle 1. Validation against independent experimental data: A model should be validated against experimental data or field observations that have not been used in the original model parameterization. We tested the XW model against multiple independent experimental data sets including electromotive force (EMF), solubility, water vapor, and water activity measurements. The results show that the XW model is accurate and valid for its intended use for predicting trace activity coefficients and therefore Nd solubility in repository environments. (2) Principle 2. Testing for relevant and sensitive variables: Solution pH is such a variable for an STM and easily acquirable. All three models are checked for their ability to predict pH conditions in Nd(NO 3 ) 3 electrolyte solutions. The OWC model fails to provide a reasonable estimate for solution pH conditions, thus casting serious doubt on its validity for a source-term calculation. In contrast, both the XW and GLC models predict close-to-neutral pH values, in agreement with experimental measurements. (3) Principle 3. Honoring physical constraints: Upon close examination, it is found that the Nd(III)-NO 3 association schema in the OWC model suffers from two shortcomings. Firstly, its second stepwise stability constant for Nd(NO 3 ) 2+ (log K 2 ) is much higher than the first stepwise stability constant for NdNO 3 2+ (log K 1 ), thus violating the general rule of (log K 2 –log K 1 ) < 0, or $\frac{K1}{K2}$>1. Secondly, the OWC model predicts abnormally high activity coefficients for Nd(NO 3 ) 2 + (up to ~900) as the concentration increases. (4) Principle 4. Minimizing degrees of freedom for model fitting: The OWC model with nine fitted parameters is compared with the GLC model with five fitted parameters, as both models apply to the concentrated region for Nd(NO 3 ) 3 electrolyte solutions. The latter appears superior to the former because the latter can fit osmotic coefficient data equally well with fewer model parameters. The work presented here thus illustrates the salient points of geochemical model development, selection, and validation in nuclear waste management.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
NASA NTRS2001

Icehouse Effect: A Selective Arctic Cooling Trend Current Models are Missing

The icehouse effect is a hypothesized climate feedback mechanism which could result in human-caused surface cooling trends in polar regions. Once understood in detail, it becomes apparent that these trends, which are discernable in the literature, but have been largely dismissed, do not conflict with the consensus assessment of the evidence, which infers century-scale Arctic warming. In fact, confirmation of the hypothesis would substantially strengthen the argument that there is a detectable human influence on today's climate. This apparent enigma is resolved only through careful attention to the detail of the hypothesis and the data supporting it. The posited surface cooling is entirely dependent on the existence of climate warming in layers capping the stable boundary layer. Also, the cooling is not pandemic, but is selective. It is readily revealed in properly sorted data by making use of the principles of micrometeorological similarity. Specifically, the cooling is manifest under a range of favorable turbulence conditions which can develop and disappear locally on time scales of minutes to hours because of the intrinsically intermittent nature of stable boundary layer turbulence. Because of the fine-scale nature of the processes which produce the cooling, modeling it is a difficult proposition. Vertical resolution on the order of 1 meter is required. Adequate models of intermittent surface fluxes coupled with radiation exchange do not currently exist, not as parameterizations for aggregated systems, nor in large eddy simulation (LES) models. This presentation will introduce the theory. An important testable null hypothesis emerges: the icehouse effect produces a unique signature or "fingerprint" which could not be produced by any other known process. The presence of this signature will be demonstrated using nearly all available Arctic temperature observations. Its aggregate effect is clearly found in Arctic monthly surface temperature trends when sorted by climatological stability. Using all available Arctic rawinsonde ascents - about 1.1 million profiles, "frozen moments" of the icehouse processes are captured in various states. Because turbulent time scales are so short in the stable boundary layer. each of these snapshots can be treated as independent -- their chronology is irrelevant. Micrometeorological similarity is invoked to reassemble the soundings into bins of similar stability and it is in a wide, coherent range of these stability bins where the cooling effect is revealed.

Wetzel, Peter J.↗
NASA NTRS1999

Selecting Meteorological Input for the Global Modeling Initiative Assessments

The Global Modeling Initiative (GMI) science team has developed a three dimensional chemistry and transport model (CTM) to evaluate the impact of the exhaust of supersonic aircraft on the stratosphere. An important goal of the GMI is to test modules for numerical transport, photochemical integration, and model dynamics within a common framework. This work is focussed on the dependence of the overall assessment on the wind and temperature fields used by the CTM. Three meteorological data sets for the stratosphere were available to GMI: the National Center for Atmospheric Research Community Climate Model (CCM2), the Goddard Earth Observing System Data Assimilation System (GEOS-DAS), and the Goddard Institute for Space Studies general circulation model (GISS-2'). Objective criteria were established by the GMI team to evaluate which of these three data sets provided the best representation of trace gases in the stratosphere today. Tracer experiments were devised to test various aspects of model transport. Stratospheric measurements of long-lived trace gases were selected as a test of the CTM transport. This presentation describes the criteria used in grading the meteorological fields and the resulting choice of wind fields to be used in the GMI assessment. This type of objective model evaluation will lead to a higher level of confidence in these assessments. We suggest that the diagnostic tests shown here be used to augment traditional general circulation model evaluation methods.

Strahan, Susan↗
NASA NTRS1993

Objective determination of image end-members in spectral mixture analysis

Spectral mixture analysis was shown to be a powerful, multifaceted tool for analysis of multi- and hyper-spectral data. The essence of the first phase of the approach is to determine a set of image end-members that best account for the spectral variance in an image cube within a constrained, linear least squares mixing model. The selection of the image end-member is usually achieved using a priori knowledge and successive trial and error solutions to refine the total number and physical location of the end-members. However, in many situations a more objective method of determining these essential components is desired. The problem of image end-member determination was approached objectively by using the inherent variance of the data. Unlike purely statistical methods such as factor analysis, this approach derives solutions that conform to a physically realistic model.

Tompkins, Stefanie↗
DOE OSTI2023

Selecting representative geological realizations to model subsurface CO 2 storage under uncertainty

Carbon capture and storage (CCS) is one of the quickest and most effective solutions for reducing carbon emissions. The majority of subsurface storage occurs in saline aquifers, for which geological information is lacking which in turn results in geological uncertainty. To evaluate uncertainty in CO 2 injection projections, the use of multiple geological realizations (GRs) has been practiced very commonly. In this approach, hundreds or thousands of high-resolution GRs is used that quickly becomes computationally expensive. This issue can be addressed with representative geological realizations (RGRs) that preserve the uncertainty domain of the ensemble GRs. Here, in this study, we propose the use of unsupervised machine learning (UML) frameworks, including dissimilarity measurement, dimensionality reduction, clustering and sampling algorithms ta select a predetermined number of RGRs. We compare the simulation outputs of the RGR sets and the ensemble using the Kolmogorov–Smirnov (KS) test to select the best UML. The UML frameworks and their associated selection processes are evaluated using a saline aquifer with a single CO 2 injection well and 200 GRs with varying uncertain petrophysical characteristics. The best UML framework is selected to use only 5% of the GRs while maintaining the uncertainty domain of the ensemble GRs. In addition, the best UML framework is tested using a saline aquifer with three CO 2 injection wells and varied GRs. The results show that our proposed UML framework can be used to choose RGRs, capturing the whole uncertainty domain. Our approach leads to a significant reduction in the computational cost associated with scenario testing, decision-making, and development planning for CO 2 storage sites under geological uncertainty.

58 GEOSCIENCES↗
NASA NTRS1998

Experiments to Determine Whether Recursive Partitioning (CART) or an Artificial Neural Network Overcomes Theoretical Limitations of Cox Proportional Hazards Regression

New computationally intensive tools for medical survival analyses include recursive partitioning (also called CART) and artificial neural networks. A challenge that remains is to better understand the behavior of these techniques in effort to know when they will be effective tools. Theoretically they may overcome limitations of the traditional multivariable survival technique, the Cox proportional hazards regression model. Experiments were designed to test whether the new tools would, in practice, overcome these limitations. Two datasets in which theory suggests CART and the neural network should outperform the Cox model were selected. The first was a published leukemia dataset manipulated to have a strong interaction that CART should detect. The second was a published cirrhosis dataset with pronounced nonlinear effects that a neural network should fit. Repeated sampling of 50 training and testing subsets was applied to each technique. The concordance index C was calculated as a measure of predictive accuracy by each technique on the testing dataset. In the interaction dataset, CART outperformed Cox (P less than 0.05) with a C improvement of 0.1 (95% Cl, 0.08 to 0.12). In the nonlinear dataset, the neural network outperformed the Cox model (P less than 0.05), but by a very slight amount (0.015). As predicted by theory, CART and the neural network were able to overcome limitations of the Cox model. Experiments like these are important to increase our understanding of when one of these new techniques will outperform the standard Cox model. Further research is necessary to predict which technique will do best a priori and to assess the magnitude of superiority.

Kattan, Michael W.↗