Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “automated modelling”

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 307 records · Page 17
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Automated Resonance Fitting for Nuclear Data Evaluation

Global and national efforts to deliver high-quality nuclear data to users have a wide-ranging impact, affecting applications in national security, reactor operations, basic science, medicine, and more. Cross section evaluation is a major part of this effort, combining theory and experimentation to produce recommended values and uncertainties for reaction probabilities. Resonance region evaluation is a specialized type of nuclear data evaluation that can require significant manual effort and months of time from expert scientists. In this article, non-convex non-linear optimization methods are combined with concepts of inferential statistics to infer a resonance model from experimental data in an automated manner that is not dependent on prior evaluation(s). This methodology aims to enhance the workflow of a resonance evaluator by minimizing time, effort, and the potential for bias from prior assumptions, while enhancing reproducibility and documentation, thereby addressing well-known challenges in the field.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2023

Advanced Tritium Process Analytics and Optimization via a Digital Twin, SRNL-TR-2023-00550

Improving the process knowledge and understanding of TCAP can be achieved by incorporating advanced tools. This project addresses “Advanced analytics for modeling, forecasting, & optimization of the tritium refinement process” and is being applied to the TCAP process. The value of digitization and advanced analytics for TCAP data are tracking of gas mixtures & inventories is presently a manual effort that is rather cumbersome. Incorporating some automation into the data capture will reduce errors and storing data in an accessible database that can be used for tracking as well as process improvements. In addition, the database approach will enable longer term history to be maintained rather than the current practice of deleting data after six months. The digitized and automated data will allow for models to be developed for the process and will enable the forecasting and optimization.

Korinko, Paul S.↗
DOE OSTI2020

Fast inference of Boosted Decision Trees in FPGAs for particle physics

We describe the implementation of Boosted Decision Trees in the hls4ml library, which allows the translation of a trained model into FPGA firmware through an automated conversion process. Thanks to its fully on-chip implementation, hls4ml performs inference of Boosted Decision Tree models with extremely low latency. With a typical latency less than 100 ns, this solution is suitable for FPGA-based real-time processing, such as in the Level-1 Trigger system of a collider experiment. These developments open up prospects for physicists to deploy BDTs in FPGAs for identifying the origin of jets, better reconstructing the energies of muons, and enabling better selection of rare signal processes.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

ON THE EFFECTIVENESS OF LLMS IN UNIT TEST GENERATION FOR STRUCTURED TEXT PROGRAMS

The reliability of industrial automation systems heavily depends on the correctness of Programmable Logic Controller (PLC) programs, which are often written in Structured Text (ST). While Large Language Models (LLMs) have shown promise in automating test generation for mainstream programming languages, their effectiveness for the syntactically strict ST language remains underexplored. This thesis presents a systematic empirical evaluation of three state-of-the-art LLMs—GPT-4o, Gemini 2.5 Pro, and Claude Sonnet 4.5—for generating ST unit tests. We examine three prompting strategies: Natural Language (NL), Code Language (CL), and Chain-of-Thought (CoT), across a curated set of 11 ST function blocks. The quality of the generated tests is assessed using Compilation Success Rate (CSR), Statement Coverage (SC), and Branch Coverage (BC). In the zero-shot setting, Claude Sonnet 4.5 achieves the highest CSR, while Gemini 2.5 Pro consistently delivers the best statement and branch coverage, particularly under CL prompts. By incorporating a one-shot CL prompt, all models exhibit substantial improvements—most notably GPT-4o, whose CSR increases from 45.45% to 90.91%, with substantial gains in both SC and BC. To further contextualize these findings, we compare GPT-4o’s one-shot results with PLCAutoTester, a state-ofthe- art ST unit test generation tool, on an additional benchmark dataset. While LLMgenerated tests approach competitive coverage levels, PLCAutoTester maintains significantly higher and more stable coverage across programs. This study provides the first comprehensive benchmark of modern LLMs for ST unit testing, highlighting their strengths, limitations, and improvements through one-shot prompting, and positioning their performance relative to specialized automated testing tools in industrial automation.

42 ENGINEERING↗
DOE OSTI2022

Sequential Decision Making (SDM) for Mesh Refinement and Model Selection in Multiscale, Multi-Physics Applications

Intelligent automation and decision support are needed to enhance computational efficiency and robustness in multiscale and multi-physics problems, including materials science, manufacturing, and climate and weather modeling. Current scientific computing approaches for enabling decisions by scientists fail to explore the role of learning, reasoning, and probabilistic planning. Often these decisions are not performed in real-time during the computation but are made prior to the start of the computation, which must be interrupted in order to make changes to the prior choices. Such interruptions at different stages of the computation increase the total computing time and the need for a human expert to frequently monitor the results. State of art scientific computing methods consist of rule-based algorithms that cannot automatically adapt to a dynamically changing computing environment. The development of a Sequential Decision Making (SDM) framework will automate scientific computing by optimizing the policies for mesh refinement, time-stepping, model and algorithm selection, resource allocation, and pre and post-processing. Our agent SDM framework for scientific computing will consist of data-driven learning (Classifier), automated reasoning (contextual knowledge), and probabilistic planning (Reinforcement Learning). In this project, we focused on three problems to demonstrate our SDM framework on a set of ordinary and partial differential equations. Classification of Lorenz system regions using Feed-Forward Neural Networks examined learning in the SDM framework. On the other hand, reasoning and planning in the SDM framework were used in two problems: adaptive time-stepping for nonlinear ODEs using on-policy RL algorithms, and adaptive mesh refinement for 2-D PDEs using off-policy RL algorithms.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

How do perceptions of safety and car ownership importance affect autonomous vehicle adoption?

In this study, we explicitly modeled how individuals' perceptions of automated vehicle (AV) safety and the importance they place on car ownership affect mode choices involving conventional and automated vehicles in the context of privately owned cars and ridehailing services. We adopted psychometric questions to capture these two latent variables and designed a stated preference survey based on the participants' actual travel patterns. Then, we quantified the impact of these latent variables on mode choices using an integrated choice and latent variable (ICLV) model. We found that both latent variables have a statistically significant effect on mode choices. The results show that car ownership importance has the most potent effect on privately owned cars (conventional car and self-driving car), followed by driverless ridehailing and conventional ridehailing. We also found that changes in safety perception are equivalent to sizable changes in price. In addition, we further investigated the impact of improvements in safety perception through four scenarios. The scenario testing results show that as the distribution of perceived safety is compressed toward positive safety perception, the market share of AVs spikes and dominates regular cars. Our results demonstrate that based on our respondents' current understanding of AVs, even if AV prices were comparable to regular cars, we cannot expect widespread use of AVs. However, improvements in AVs' safety and, consequently, consumer safety perception can considerably expand AVs' market share, and may offset the high cost of using the technology.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2023

Automated Quantification of Wind Turbine Blade Leading Edge Erosion from Field Images

Wind turbine blade leading edge erosion is a major source of power production loss and early detection benefits optimization of repair strategies. Two machine learning (ML) models are developed and evaluated for automated quantification of the areal extent, morphology and nature (deep, shallow) of damage from field images. The supervised ML model employs convolutional neural networks (CNN) and learns features (specific types of damage) present in an annotated set of training images. The unsupervised approach aggregates pixel intensity thresholding with calculation of pixel-by-pixel shadow ratio (PTS) to independently identify features within images. The models are developed and tested using a dataset of 140 field images. The images sample across a range of blade orientation, aspect ratio, lighting and resolution. Each model (CNN v PTS) is applied to quantify the percent area of the visible blade that is damaged and classifies the damage into deep or shallow using only the images as input. Both models successfully identify approximately 65% of total damage area in the independent images, and both perform better at quantifying deep damage. The CNN is more successful at identifying shallow damage and exhibits better performance when applied to the images after they are preprocessed to a common blade orientation.

Aird, Jeanie A.↗
DOE OSTI2025

32 examples of LLM applications in materials science and chemistry: towards automation, assistants, agents, and accelerated scientific discovery

Abstract Large language models (LLMs) are reshaping many aspects of materials science and chemistry research, enabling advances in molecular property prediction, materials design, scientific automation, knowledge extraction, and more. Recent developments demonstrate that the latest class of models are able to integrate structured and unstructured data, assist in hypothesis generation, and streamline research workflows. To explore the frontier of LLM capabilities across the research lifecycle, we review applications of LLMs through 32 total projects developed during the second annual LLM hackathon for applications in materials science and chemistry, a global hybrid event. These projects spanned seven key research areas: (1) molecular and material property prediction, (2) molecular and material design, (3) automation and novel interfaces, (4) scientific communication and education, (5) research data management and automation, (6) hypothesis generation and evaluation, and (7) knowledge extraction and reasoning from the scientific literature. Collectively, these applications illustrate how LLMs serve as versatile predictive models, platforms for rapid prototyping of domain-specific tools, and much more. In particular, improvements in both open source and proprietary LLM performance through the addition of reasoning, additional training data, and new techniques have expanded effectiveness, particularly in low-data environments and interdisciplinary research. As LLMs continue to improve, their integration into scientific workflows presents both new opportunities and new challenges, requiring ongoing exploration, continued refinement, and further research to address reliability, interpretability, and reproducibility.

Computer Science↗
DOE OSTI2022

BFSVBF (BatFinder Smart Video BioFilter) [SWR-22-87] and Multi-class BatFinder Smart Video BioFilter Keras

Bats are notoriously difficult to study, therefore, identifying specific behavioral trends and the precise environmental conditions at the time of collision requires a monitoring solution that can reliably collect relevant data. To date, thermal infrared video surveillance has been extensively applied to study bats and has proven to be a powerful yet cumbersome tool. Current analytical approaches are time consuming because data processing data has not been fully automated. In the past, steps have been taken to record avian and bat activity in conjunction with complicated image processing techniques that separate species from other moving objects within the field of view (i.e. clouds and portions of the wind turbine). Once the videos are collected, the post-processing does not allow real time monitoring and identification, leading to a delay in both studying the behavior of these species and determining the effectiveness of any impact reduction strategy being studied. Moreover, object identification capability is lacking, thus limiting the usefulness of video data. To resolve these issues, we are using open source computer vision and machine learning techniques allowing for automatic detection of objects in real-time with the ability to correlate these objects with environmental variables and recording the flight paths of each object. The code has gone through five rounds of development with images used to train the models. This advancement allows for automated real-time data collection, identification, and tracking, thereby eliminating the need for long and tedious post-analysis processing of the videos. We will discuss the two open source and publicly available machine learning models developed within this scope of this work: 1) a binary model with a 97.5% accuracy in identifying the difference between an object and an empty scene, including wind turbine and clouds; and 2) a multiple classification model with the capability of identifying the type of object detected: bats (90% accuracy), birds (83% accuracy), insects (69% accuracy) and non-biological (99% accuracy).

Yarbrough, John↗
DOE OSTI2025

SIDDA: SInkhorn Dynamic Domain Adaptation for image classification with equivariant neural networks

Modern neural networks (NNs) often do not generalize well in the presence of a ‘covariate shift’; that is, in situations where the training and test data distributions differ, but the conditional distribution of classification labels given the data remains unchanged. In such cases, NN generalization can be reduced to a problem of learning more robust, domain-invariant features. Domain adaptation (DA) methods include a broad range of techniques aimed at achieving this; however, these methods have struggled with the need for extensive hyperparameter tuning, which then incurs significant computational costs. In this work, we introduce SInkhorn Dynamic Domain Adaptation (SIDDA), an out-of-the-box DA training algorithm built upon the Sinkhorn divergence, that can achieve effective domain alignment with minimal hyperparameter tuning and computational overhead. We demonstrate the efficacy of our method on multiple simulated and real datasets of varying complexity, including simple shapes, handwritten digits, real astronomical observations, and remote sensing data. These datasets exhibit covariate shifts due to noise, blurring, differences between telescopes, and variations in imaging wavelengths. SIDDA is compatible with a variety of NN architectures, and it works particularly well in improving classification accuracy and model calibration when paired with symmetry-aware equivariant NNs (ENNs). We find that SIDDA consistently enhances the generalization capabilities of NNs, achieving up to a ${\approx}40\%$ improvement in classification accuracy on unlabeled target data, while also providing a more modest performance gain of $\lesssim 1\%$ on labeled source data. We also study the efficacy of DA on ENNs with respect to the varying group orders of the dihedral group DN, and find that the model performance improves as the degree of equivariance increases. Finally, if SIDDA achieves proper domain alignment, it also enhances model calibration on both source and target data, with the most significant gains in the unlabeled target domain—achieving over an order of magnitude improvement in the expected calibration error and Brier score. SIDDA’s versatility across various NN models and datasets, combined with its automated approach to domain alignment, has the potential to significantly advance multi-dataset studies by enabling the development of highly generalizable models.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Development and Validation of Home Comfort System for Total Performance Deficiency/Fault Detection and Optimal Comfort Control

In this project, we developed and tested a learning-based home thermal model that facilitates the operation of a model predictive control (MPC)-based optimization agent and an automated fault detection and diagnosis (AFDD) agent. The home thermal model was constructed using a two-node resistor-capacitor model. Moreover, two accompanying parameter identification methods were introduced, least-squares and optimization. Based on the home thermal model, the MPC-based optimization agent was developed to optimize residential HVAC operation. Using two FDD methods, the AFDD agent was constructed to detect and diagnose two prevalent residential AC faults, airflow reduction and refrigerant undercharge. The home thermal model, along with the MPC-based optimization agent and AFDD agent, were tested at the Norman Test House, Miami Test House, Pacific Northwest National Laboratory (PNNL) Test House A, and PNNL Test House B. Finally, they were also field tested in nine demonstration homes with real occupants.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2023

Effect of image resolution on automated classification of chest X-rays

Deep learning (DL) models have received much attention lately for their ability to achieve expert-level performance on the accurate automated analysis of chest X-rays (CXRs). Recently available public CXR datasets include high resolution images, but state-of-the-art models are trained on reduced size images due to limitations on graphics processing unit memory and training time. As computing hardware continues to advance, it has become feasible to train deep convolutional neural networks on high-resolution images without sacrificing detail by downscaling. This study examines the effect of increased resolution on CXR classification performance. We used the publicly available MIMIC-CXR-JPG dataset, comprising 377,110 high resolution CXR images for this study. We applied image downscaling from native resolution to 2048 × 2048 pixels, 1024 × 1024 pixels, 512 × 512 pixels, and 256 × 256 pixels and then we used the DenseNet121 and EfficientNet-B4 DL models to evaluate clinical task performance using these four downscaled image resolutions. We find that while some clinical findings are more reliably labeled using high resolutions, many other findings are actually labeled better using downscaled inputs. We qualitatively verify that tasks requiring a large receptive field are better suited to downscaled low resolution input images, by inspecting effective receptive fields and class activation maps of trained models. Lastly, we show that stacking an ensemble across resolutions outperforms each individual learner at all input resolutions while providing interpretable scale weights, indicating that diverse information is extracted across resolutions.

47 OTHER INSTRUMENTATION↗
DOE OSTI2023

Using CUBIT to Create Unstructured Mesh Models for MCNP Simulations

The Monte Carlo N-Particle (MCNP) transport code version 6 (also known as MCNP6) has the capability for tracking particles on unstructured mesh (UM) geometry models embedded into constructive solid geometry (CSG) cells. This feature has been developed for performing calculations of complex geometry models because creating CSG models is a time-consuming and error-prone process as the complexities of geometries increase. An MCNP UM calculation requires UM geometry input files. The UM capability was originally designed to work with UM models created with the Abaqus/CAE software and ASCII input files that it generates. The Abaqus-formatted input files needed for MCNP UM calculations must have the correct Abaqus syntax and meet the additional MCNP requirements. Several software packages can generate a UM model formatted as an Abaqus input file. Cubit, the Sandia National Laboratory automated mesh generation toolkit, can generate a UM model formatted as an Abaqus input file. However, the Abaqus input files created by Cubit cannot be used for MCNP simulations. A Python script has been developed to convert an Abaqus file created by Cubit to an Abaqus file that the MCNP code can process. This report describes the process of using Cubit to create UM models for MCNP calculations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Assessment of Technoeconomic Opportunities in Automation for Nuclear Microreactors

Achieving full decarbonization of all economic sectors remains a challenge, especially in niche markets. For example, remote communities and industrial or mining activities detached from the main electric grid heavily rely on fossil fuels, similar to urban and industrial microgrids with combined heat and power needs. A combination of renewables and energy storage is often not suitable due to cost, reliability, intermittency, and large storage requirements. Small nuclear reactors with a flexible purpose could serve these applications. Microreactors (MR) are a class of reactors that are compact, factory manufactured, transportable, and self-regulating. Typically, they generate much less power than their large reactor counterparts. The main advantages of microreactors include the versatile nature of the energy produced, the reliability of supply, and freedom from having to transport and store large quantities of fuels on-site, coupled with the absence of dependence on an electrical grid. A strong business case is needed to move from the microreactor prototype to the commercialization phase. In fact, fossil fuels are still relatively inexpensive, and in the near term, carbon credits will be available to virtually compensate for emissions. For microreactors, one of the main costs in operation and maintenance (O&M) is their staffing levels. In this study, we investigate how to optimize the number (and thus the cost) of workers, moving from a traditional, fully manned, on-site personnel approach to an unmanned, remote personnel approach. We examine four different staffing models that can be implemented as the technology matures and evolves. We estimate the staffing needs of each model and build a business case to justify the substitution of on-site personnel with adequate technologies. To do so, we propose a cost model to quantify potential cost reductions from automating O&M activities. The model accounts for both the reduction in cost derived from the reduced number of full-time-equivalent (FTE) employees and the increase in cost derived from the need to buy new control hardware as needed. Applying the cost model that we created to different scenarios, an on-site O&M cost reduction exceeding 80% can be expected. Additionally, we found that it is more impactful to focus on automating routine O&M tasks rather than attempting to automate transient management (shutdowns, restarts, monitoring condition deviations). In fact, transients typically account for less than 1% of the total FTE time spent on the reactors.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2021

Automating Bayesian inference and design to quantify acoustic particle levitation

Self-propulsion of micro- and nanoparticles powered by ultrasound provides an attractive strategy for the remote manipulation of colloidal matter using biocompatible energy inputs. Quantitative understanding of particle motion and its dependence on size, shape, and composition requires accurate characterization of the acoustic field, which depends sensitively on the experimental setup. Here, we show how automated experiments based on Bayesian inference and design can accurately and efficiently characterize the acoustic field within resonant chambers used to propel acoustic nanomotors. Repeated cycles of observation, inference, and design (OID) are guided by a physical model that describes the rate at which levitating particles approach the nodal plane. Using video microscopy, we observe the relaxation of tracer particles to this plane following the application of the acoustic field. We use sequential Monte Carlo methods to infer model parameters such as the amplitude and frequency of the resonant chamber while accounting for particle-level measurement noise and population-level heterogeneity in the field. Guided by simulated outcomes, we select the optimal design for the next experiment as to maximize the information gain in the relevant parameters. We show how this iterative process serves to discriminate between competing hypotheses and efficiently converges to accurate parameter estimates using only few automated experiments. We discuss the need for model criticism to ensure the validity of the guiding model throughout automated cycles of observation, inference, and design. Furthermore, this work demonstrates how Bayesian methods can learn the parameters of nonlinear, hierarchical models used to describe video microscopy data of active colloids.

42 ENGINEERING↗
DOE OSTI2023

Natural Language Processing for Text Based Event Extraction: Identifying Events of Interest Related to Worldwide State-Sponsored Civil Nuclear Power

Beginning in FY20, SRNL was funded by the National Nuclear Security Administration’s Office of Defense Nuclear Non-Proliferation Research and Development to develop a prototype natural language processing/natural language understating machine learning-based modeling and analysis pipeline to extract and forecast events of interest from massive open data sources. The working hypothesis within the approach is that contextual shifts in key words and phrases act as indicators of events of interest over time. Therefore, by identifying points in time where contextual shifts occur, events of interest can be extracted along with explicit and implicit connections of entities and activities. The development of the preliminary prototype pipeline proved successful, meriting further testing of the pipeline on more broad topical domains and in a worldwide data environment. Therefore, SRNL, in collaboration with the Sanghani Center for Artificial Intelligence and Data Analytics at Virginia Tech, have continued development with a test case of identifying events of interest related to worldwide state-sponsored civil nuclear power in open data sources. In the first year of this follow-on effort, the team has curated domain-specific data corpuses using an automated scheme and applied the modeling and analysis pipeline. This robust, focused, and efficient approach consists of an ensemble of analyses applied to time dependent word embedding models that are trained on the data corpuses. In this report, the team has demonstrated the capability of the existing pipeline (as development has continued in parallel) by exploring several specific case-studies centered around Rosatom’s international activities regarding the planning, construction, operation, and/or shutdown of nuclear reactors. A basic timeline events has been generated by manually cataloging known “milestone” events that have occurred at reactors in Turkey, Finland, Hungary, and Egypt and compared with the output of the modeling pipeline. In this approach, the team has characterized the lead time using the prototype pipeline, as well as the ability to capture relevant information, which proved 100% successful. A deep dive example of the Akkuyu reactor (Turkey) is presented that shows the breadth of information that can be captured using the approach. In this case study, events were extracted pertaining to the planning/construction of Akkuyu including protests from the population, information campaigns in response to the protests, forged regulatory documents and lawsuits, budgetary/shareholder information, geopolitical tensions, and the various construction milestones. This has demonstrated the pipeline’s utility as a research aid or real-time event extraction tool, where summary-level information and detailed text extractions from millions of articles or Tweets across long time periods can be generated with significantly less effort than current techniques.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2021

Rapid Construction of Validated Chemistry Models for Advanced Biofuels (Co-Optima Final Report)

Computer methods were developed and demonstrated that make it possible to predict the combustion and pyrolysis chemistry of proposed biofuels, even before the biofuels are ever manufactured or experimentally tested. The methods are based on first-principles calculations of reaction rates and equilibria. A large number of complicated calculations are required, so we developed a computer workflow that automates many of the steps. With the chemical structure of the biofuel and the system conditions (temperature, pressure, compositions, etc.) as the input, the computer methods can construct a detailed kinetic model with refined parameters without intense human intervention. The automation of model construction makes it practical for a single engineer to make the predictions in a reasonable time period. The accuracy of the computer predictions was tested by measuring the time-profiles of various reaction intermediates (carbon monoxide, carbon dioxide, water, formaldehyde, methane, and ethylene) formed as the biofuels pyrolyze or ignite using sophisticated laser shock tube methods. Several biofuels, including methyl propyl ether, cyclopentanone, butyl acetate isomers, etc., were carefully studied to demonstrate the validity of the proposed computational and experimental methodology. This research project opens up the possibility for assessing some proposed biofuels on the computer without expensive experimental campaigns. The new computer methods are open source and well-documented, allowing a knowledgeable modeler to easily access the developed methods and implement them in their work routine. Moreover, they are also convenient and powerful, promising for developing the fuel chemistry models needed for use in engine simulators, facilitating co-optimization of new engine designs with new fuels.

09 BIOMASS FUELS↗
DOE OSTI2023

Using Ultrasound Image Augmentation and Ensemble Predictions to Prevent Machine-Learning Model Overfitting

Deep learning predictive models have the potential to simplify and automate medical imaging diagnostics by lowering the skill threshold for image interpretation. However, this requires predictive models that are generalized to handle subject variability as seen clinically. Here, we highlight methods to improve test accuracy of an image classifier model for shrapnel identification using tissue phantom image sets. Using a previously developed image classifier neural network—termed ShrapML—blind test accuracy was less than 70% and was variable depending on the training/test data setup, as determined by a leave one subject out (LOSO) holdout methodology. Introduction of affine transformations for image augmentation or MixUp methodologies to generate additional training sets improved model performance and overall accuracy improved to 75%. Further improvements were made by aggregating predictions across five LOSO holdouts. This was done by bagging confidences or predictions from all LOSOs or the top-3 LOSO confidence models for each image prediction. Top-3 LOSO confidence bagging performed best, with test accuracy improved to greater than 85% accuracy for two different blind tissue phantoms. This was confirmed by gradient-weighted class activation mapping to highlight that the image classifier was tracking shrapnel in the image sets. Overall, data augmentation and ensemble prediction approaches were suitable for creating more generalized predictive models for ultrasound image analysis, a critical step for real-time diagnostic deployment.

60 APPLIED LIFE SCIENCES↗