Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “decision tree”

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 325 records · Page 18
All sourcesNASA NTRSDOE OSTI
DOE OSTI2013

Gunnar uranium mine environmental remediation - Northern Saskatchewan

The Gunnar Mine and mill site was the largest of some 38 now-abandoned uranium mines that were developed and operated in Northern Saskatchewan, Canada, during the Cold War years. During their operating lifetimes these mines produced large quantities of ore and tailings. The Gunnar mine (open pit and underground) produced over 5 million tonnes of uranium ore and nearly 4.4 million tonnes of mine tailings during its operations from 1955 through 1963. An estimated 2.2 to 2.7 million m{sup 3} of waste rock that was generated during the processing of the ore abuts the shores of Lake Athabasca, the 22. largest lake in the world. After closure in the 1960's, the Gunnar site was abandoned with little to no decommissioning being done. The Saskatchewan Research Council has been contracted to manage the clean-up of these abandoned northern uranium mine and mill sites. The Gunnar Mine, because of the magnitude of tailings and waste rock, is subject to an environmental site assessment process regulated by both provincial and federal governments. This process requires a detailed study of the environmental impacts that have resulted from the mining activities and an analysis of projected impacts from remediation efforts. The environmental assessment process, specific site studies, and public involvement initiatives are all now well underway. Due to the many uncertainties associated with an abandoned site, an adaptive remediation approach, utilizing a decision tree, presented within the environmental assessment documents will be used as part of the site regulatory licensing. A critical early task was dealing with major public safety hazards on the site. The site originally included many buildings that were remnants of a community of approximately 800 people who once occupied the site. These buildings, many of which contained high levels of asbestos, had to be appropriately abated and demolished. Similarly, the original mine head frame and mill site buildings, many of which still contained the original machinery and equipment, also had to be dismantled. Remediation options for the accumulated demolition debris have been assessed, as have remediation options for the waste rock and tailings, all of which form part of the environmental assessment. The regulatory requirements include the environmental assessment processes, a complex public involvement strategy, and licensing from the Canadian Nuclear Safety Commission (CNSC) with the long-term goal of releasing the property in a remediated and stable state to the Province of Saskatchewan. Prescribed environmental and land use endpoints will be determined based on the environmental assessment studies and remediation options analyzed and implemented. Ultimately, the site will be released into an institutional controls program that will allow long-term government management and monitoring. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2016

TU-FG-201-11: Evaluating the Validity of Prospective Risk Analysis Methods: A Comparison of Traditional FMEA and Modified Healthcare FMEA

Purpose: To examine the ability of traditional Failure mode and effects analysis (FMEA) and a light version of Healthcare FMEA (HFMEA), called Scenario analysis of FMEA (SAFER) by comparing their outputs in terms of the risks identified and their severity rankings. Methods: We applied two prospective methods of the quality management to surface image guided, linac-based radiosurgery (SIG-RS). For the traditional FMEA, decisions on how to improve an operation are based on risk priority number (RPN). RPN is a product of three indices: occurrence, severity and detectability. The SAFER approach; utilized two indices-frequency and severity-which were defined by a multidisciplinary team. A criticality matrix was divided into 4 categories; very low, low, high and very high. For high risk events, an additional evaluation was performed. Based upon the criticality of the process, it was decided if additional safety measures were needed and what they comprise. Results: Two methods were independently compared to determine if the results and rated risks were matching or not. Our results showed an agreement of 67% between FMEA and SAFER approaches for the 15 riskiest SIG-specific failure modes. The main differences between the two approaches were the distribution of the values and the failure modes (No.52, 54, 154) that have high SAFER scores do not necessarily have high FMEA RPN scores. In our results, there were additional risks identified by both methods with little correspondence. In the SAFER, when the risk score is determined, the basis of the established decision tree or the failure mode should be more investigated. Conclusion: The FMEA method takes into account the probability that an error passes without being detected. SAFER is inductive because it requires the identification of the consequences from causes, and semi-quantitative since it allow the prioritization of risks and mitigation measures, and thus is perfectly applicable to clinical parts of radiotherapy.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2023

A bacterial sensor taxonomy across earth ecosystems for machine learning applications

Microbial communities have evolved to colonize all ecosystems of the planet, from the deep sea to the human gut. Microbes survive by sensing, responding, and adapting to immediate environmental cues. This process is driven by signal transduction proteins such as histidine kinases, which use their sensing domains to bind or otherwise detect environmental cues and “transduce” signals to adjust internal processes. We hypothesized that an ecosystem’s unique stimuli leave a sensor “fingerprint,” able to identify and shed insight on ecosystem conditions. To test this, we collected 20,712 publicly available metagenomes from Host-associated, Environmental, and Engineered ecosystems across the globe. We extracted and clustered the collection’s nearly 18M unique sensory domains into 113,712 similar groupings with MMseqs2. We built gradient-boosted decision tree machine learning models and found we could classify the ecosystem type (accuracy: 87%) and predict the levels of different physical parameters (R2 score: 83%) using the sensor cluster abundance as features. Feature importance enables identification of the most predictive sensors to differentiate between ecosystems which can lead to mechanistic interpretations if the sensor domains are well annotated. To demonstrate this, a machine learning model was trained to predict patient’s disease state and used to identify domains related to oxygen sensing present in a healthy gut but missing in patients with abnormal conditions. Moreover, since 98.7% of identified sensor domains are uncharacterized, importance ranking can be used to prioritize sensors to determine what ecosystem function they may be sensing. Furthermore, these new predictive sensors can function as targets for novel sensor engineering with applications in biotechnology, ecosystem maintenance, and medicine.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2018

Machine Learning Leveraging Genomes from Metagenomes Identifies Influential Antibiotic Resistance Genes in the Infant Gut Microbiome

Antibiotic resistance in pathogens is extensively studied, and yet little is known about how antibiotic resistance genes of typical gut bacteria influence microbiome dynamics. Here, we leveraged genomes from metagenomes to investigate how genes of the premature infant gut resistome correspond to the ability of bacteria to survive under certain environmental and clinical conditions. We found that formula feeding impacts the resistome. Random forest models corroborated by statistical tests revealed that the gut resistome of formula-fed infants is enriched in class D beta-lactamase genes. Interestingly, Clostridium difficile strains harboring this gene are at higher abundance in formula-fed infants than C. difficile strains lacking this gene. Organisms with genes for major facilitator superfamily drug efflux pumps have higher replication rates under all conditions, even in the absence of antibiotic therapy. Using a machine learning approach, we identified genes that are predictive of an organism’s direction of change in relative abundance after administration of vancomycin and cephalosporin antibiotics. The most accurate results were obtained by reducing annotated genomic data to five principal components classified by boosted decision trees. Among the genes involved in predicting whether an organism increased in relative abundance after treatment are those that encode subclass B2 beta-lactamases and transcriptional regulators of vancomycin resistance. This demonstrates that machine learning applied to genome-resolved metagenomics data can identify key genes for survival after antibiotics treatment and predict how organisms in the gut microbiome will respond to antibiotic administration.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Pulse shape discrimination technique for diffuse supernova neutrino background search with JUNO

Pulse shape discrimination (PSD) is widely used in particle and nuclear physics. Specifically in liquid scintillator detectors, PSD facilitates the classification of different particle types based on their energy deposition patterns. This technique is particularly valuable for studies of the diffuse supernova neutrino background (DSNB), nucleon decay, and dark matter searches. This paper presents a detailed investigation of the PSD technique, applied in the DSNB search performed with the Jiangmen Underground Neutrino Observatory (JUNO). Instead of using conventional cut-and-count methods, we employ methods based on boosted decision trees and neural networks and compare their capability to distinguish the DSNB signals from the atmospheric neutrino neutral-current background events. The two methods demonstrate comparable performance, resulting in a 50–80% improvement in signal efficiency compared to a previous study performed for JUNO (An et al. [JUNO] in J Phys G 43(3):030401, 2016). Moreover, we study the dependence of the PSD performance on the visible energy and final state composition of the events and find a significant dependence on the presence/absence of 11 C. Finally, we evaluate the impact of the detector effects (photon propagation, PMT dark noise, and waveform reconstruction) on the PSD performance.

FOS: Physical sciences↗
DOE OSTI2025

Reconstruction and identification of pairs of collimated τ-leptons decaying hadronically using s=13 TeV pp collision data with the ATLAS detector

This paper describes an algorithm for reconstructing and identifying a highly collimated hadronically decaying τ$$\tau $$-lepton pair with low transverse momentum. When two τ$$\tau $$-leptons are highly collimated, their visible decay products might overlap, degrading the reconstruction performance for each of the τ$$\tau $$-leptons. A dedicated treatment attempting to tag the τ$$\tau $$-lepton pair as a single object is required. The reconstruction algorithm is based on a large radius jet and its associated two leading subjets, and the identification uses a boosted decision tree to discriminate between signatures from τ+τ-$$\tau ^+\tau ^-$$ systems and those arising from QCD jets. The efficiency of the identification algorithm is measured in Zγ$$Z\gamma $$ events using proton–proton collision data at s=13$$\sqrt{s}=13$$ TeV collected by the ATLAS experiment at the Large Hadron Collider between 2015 and 2018, corresponding to an integrated luminosity of 139fb-1$$139\,\text{ fb}^{-1}$$. The resulting data-to-simulation scale factors are close to unity with uncertainties ranging from 26 to 37%.

Aad, G↗
DOE OSTI2022

Measuring the electron Yukawa coupling via resonant s-channel Higgs production at FCC-ee

The Future Circular Collider (FCC-ee) offers the unique opportunity of studying the Higgs Yukawa coupling to the electron, \(y_\mathrm {e}\) , via resonant s -channel production, \(\mathrm {e^+e^-}\rightarrow \mathrm {H}\) , in a dedicated run at \(\sqrt{s} = m_\mathrm {H}\) . The signature for direct Higgs production is a small rise in the cross sections for particular final states, consistent with Higgs decays, over the expectations for their occurrence due to Standard Model (SM) background processes involving \(\mathrm {Z}^*\) , \(\gamma ^*\) , or t -channel exchanges alone. Performing such a measurement is remarkably challenging for four main reasons. First, the low value of the e \(^\pm \) mass leads to a tiny \(y_\mathrm {e}\) coupling and correspondingly small cross section: \(\sigma _\mathrm {ee\rightarrow H} \propto m_\mathrm {e}^2 = 0.57\) fb accounting for initial-state \(\gamma \) radiation. Second, the \(\mathrm {e^+e^-}\) beams must be monochromatized such that the spread of their centre-of-mass (c.m.) energy is commensurate with the narrow width of the SM Higgs boson, \(\varGamma _\mathrm {H} = 4.1\) MeV, while keeping large beam luminosities. Third, the Higgs mass must also be known beforehand with a few-MeV accuracy in order to operate the collider at the resonance peak, \(\sqrt{s} = m_\mathrm {H}\) . Last but not least, the cross sections of the background processes are many orders-of-magnitude larger than those of the Higgs decay signals. A preliminary generator-level study of 11 Higgs decay channels using a multivariate analysis, which exploits boosted decision trees to discriminate signal and background events, identifies two final states as the most promising ones in terms of statistical significance: \(\mathrm {H}\rightarrow gg\) and \(\mathrm {H}\rightarrow \mathrm {W}\mathrm {W}^*\!\rightarrow \ell \nu \) + 2 jets. For a benchmark monochromatization with 4.1-MeV c.m. energy spread (leading to \(\sigma _\mathrm {ee\rightarrow H} = 0.28\) fb) and 10 ab \(^{-1}\) of integrated luminosity, a \(1.3\sigma \) signal significance can be reached, corresponding to an upper limit on the e \(^\pm \) Yukawa coupling at 1.6 times the SM value: |y e | < 1.6|y e SM | at 95% confidence level, per FCC-ee interaction point per year. Directions for future improvements of the study are outlined.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2021

Machine learning-based analysis of COVID-19 pandemic impact on US research networks

Here in this study we explore how fallout from the changing public health policy around COVID-19 has changed how researchers access and process their science experiments. Using a combination of techniques from statistical analysis and machine learning, we conduct a retrospective analysis of historical network data for a period around the stay-at-home orders that took place in March 2020. Our analysis takes data from the entire ESnet infrastructure to explore DOE high-performance computing (HPC) resources at OLCF, ALCF, and NERSC, as well as User sites such as PNNL and JLAB. We look at detecting and quantifying changes in site activity using a combination of t-Distributed Stochastic Neighbor Embedding (t-SNE) and decision tree analysis. Our findings bring insights into the working patterns and impact on data volume movements, particularly during late-night hours and weekends.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Accelerating FEM-Based Corrosion Predictions Using Machine Learning

Atmospheric corrosion of metallic parts is a widespread materials degradation phenomena that is challenging to predict given its dependence on many factors (e.g. environmental, physiochemical, and part geometry). For materials with long expected service lives, accurately predicting the degree to which corrosion will degrade part performance is especially difficult due to the stochastic nature of corrosion damage spread across years or decades of service. The Finite Element Method (FEM) is a computational technique capable of providing accurate estimates of corrosion rate by numerically solving complex differential Eqs. characterizing this phenomena. Nevertheless, given the iterative nature of FEM and the computational expense required to solve these complex equations, FEM is ill-equipped for an efficient exploration of the design space to identify factors that accelerate or deter corrosion, despite its accuracy. In this work, a machine learning based surrogate model capable of providing accurate predictions of corrosion with significant computational savings is introduced. Specifically, this work leverages AdaBoosted Decision trees to provide an accurate estimate of corrosion current per width given different values of temperature, water layer thickness, molarity of the solution, and the length of the cathode for a galvanic couple of aluminum and stainless steel.

36 MATERIALS SCIENCE↗
DOE OSTI2018

Deep Jointly-Informed neural networks

OJINN Is an easy-to-use deep neural network model that requires fewer user-specified hyper-parameters than traditional neural networks. The algorithm leverages decision trees trained on the data to determine an appropriate deep neural network architectures and weight Initializations. Optional functions also select the learning rate, batch size. and number of training Iterations necessary to create an accurate model.

Humbird, Kelli↗
DOE OSTI2022

IViz-OT (Intrusion Detection Visualizer for Operational Technology Network) [SWR-22-63]

The Visualizer dashboard provides grid operator highly-trusted alarming environment for an ongoing or potential cyber-attack based on system anomalies and network-based verification. Once anomalies are detected by the IDS tool (HIDES, NREL SWR-19-65), this platform stores the signatures or alert logs that are generated by the intrusion detector, lays out the detailed summary of the possible alerts, and maps these attacks with high-level scenarios. These scenarios are later combined to define a final event using a decision tree approach and a final report is generated out of this tool for further forensic analysis. It also supports authentication and authorization to support roles-based access control (RBAC) for users and a group of people.

Singh, Vivek Kumar↗
DOE OSTI2025

Decayheatml

This code is designed to predict and analyze the decay heat generated in molten salt reactors (MSRs) using a hybrid approach that combines machine learning and segmented polynomial fitting. The accurate prediction of decay heat is essential for reactor safety and the optimization of spent fuel storage. The code operates through several key components: 1) Data Architecture: It incorporates a modular data architecture that handles various MSR-specific operational parameters such as power density, humidity content, and air ingress. These parameters are sampled using Sobol sequences to ensure comprehensive coverage of operational uncertainties. 2) Machine Learning Framework: The code employs a diverse set of machine learning models, including polynomial regression, decision trees, random forests, gradient boosting, support vector regression, k-nearest neighbors, multi-layer perceptrons, and symbolic regression. These models are trained to predict decay heat over a wide temporal range, from immediate shutdown up to 10,000 years. 3) Region-Optimized Training: The temporal domain is divided into multiple regions, each modeled separately to capture distinct decay heat characteristics across different time scales. This approach significantly improves the accuracy and interpretability of predictions. 4) Segmented Polynomial Interpretation (SPI): The SPI method translates machine learning predictions into piecewise polynomial equations. These equations are physically interpretable and can be directly integrated into existing engineering workflows and safety analyses. 5) Front-End Interfaces: The code includes both a Jupyter notebook interface for research development and a Streamlit web application for operational deployment. These interfaces allow users to interactively explore decay heat predictions, adjust operational parameters, and visualize results in real-time. 6) Applications: The framework supports various applications, including safety system validation and spent fuel container optimization. It enables real-time evaluation of worst-case decay heat scenarios, informing the design of passive safety systems and optimizing container designs for long-term storage. Overall, this code provides a robust, accurate, and user-friendly tool for predicting decay heat in MSRs, enhancing reactor safety, and optimizing spent fuel management.

Retamales, Mauricio Eduardo Tano [Idaho National L↗
DOE OSTI2012

A Maximum-Entropy approach for accurate document annotation in the biomedical domain

The increasing number of scientific literature on the Web and the absence of efficient tools used for classifying and searching the documents are the two most important factors that influence the speed of the search and the quality of the results. Previous studies have shown that the usage of ontologies makes it possible to process document and query information at the semantic level, which greatly improves the search for the relevant information and makes one step further towards the Semantic Web. A fundamental step in these approaches is the annotation of documents with ontology concepts, which can also be seen as a classification task. In this paper we address this issue for the biomedical domain and present a new automated and robust method, based on a Maximum Entropy approach, for annotating biomedical literature documents with terms from the Medical Subject Headings (MeSH). The experimental evaluation shows that the suggested Maximum Entropy approach for annotating biomedical documents with MeSH terms is highly accurate, robust to the ambiguity of terms, and can provide very good performance even when a very small number of training documents is used. More precisely, we show that the proposed algorithm obtained an average F-measure of 92.4% (precision 99.41%, recall 86.77%) for the full range of the explored terms (4,078 MeSH terms), and that the algorithm’s performance is resilient to terms’ ambiguity, achieving an average F-measure of 92.42% (precision 99.32%, recall 86.87%) in the explored MeSH terms which were found to be ambiguous according to the Unified Medical Language System (UMLS) thesaurus. Finally, we compared the results of the suggested methodology with a Naive Bayes and a Decision Trees classification approach, and we show that the Maximum Entropy based approach performed with higher F-Measure in both ambiguous and monosemous MeSH terms.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

Learning curves for drug response prediction in cancer cell lines

Motivated by the size and availability of cell line drug sensitivity data, researchers have been developing machine learning (ML) models for predicting drug response to advance cancer treatment. As drug sensitivity studies continue generating drug response data, a common question is whether the generalization performance of existing prediction models can be further improved with more training data. We utilize empirical learning curves for evaluating and comparing the data scaling properties of two neural networks (NNs) and two gradient boosting decision tree (GBDT) models trained on four cell line drug screening datasets. The learning curves are accurately fitted to a power law model, providing a framework for assessing the data scaling behavior of these models. The curves demonstrate that no single model dominates in terms of prediction performance across all datasets and training sizes, thus suggesting that the actual shape of these curves depends on the unique pair of an ML model and a dataset. The multi-input NN (mNN), in which gene expressions of cancer cells and molecular drug descriptors are input into separate subnetworks, outperforms a single-input NN (sNN), where the cell and drug features are concatenated for the input layer. In contrast, a GBDT with hyperparameter tuning exhibits superior performance as compared with both NNs at the lower range of training set sizes for two of the tested datasets, whereas the mNN consistently performs better at the higher range of training sizes. Moreover, the trajectory of the curves suggests that increasing the sample size is expected to further improve prediction scores of both NNs. These observations demonstrate the benefit of using learning curves to evaluate prediction models, providing a broader perspective on the overall data scaling characteristics. A fitted power law learning curve provides a forward-looking metric for analyzing prediction performance and can serve as a co-design tool to guide experimental biologists and computational scientists in the design of future experiments in prospective research studies.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

Human limits in machine learning: prediction of potato yield and disease using soil microbiome data

Abstract Background The preservation of soil health is a critical challenge in the 21st century due to its significant impact on agriculture, human health, and biodiversity. We provide one of the first comprehensive investigations into the predictive potential of machine learning models for understanding the connections between soil and biological phenotypes. We investigate an integrative framework performing accurate machine learning-based prediction of plant performance from biological, chemical, and physical properties of the soil via two models: random forest and Bayesian neural network. Results Prediction improves when we add environmental features, such as soil properties and microbial density, along with microbiome data. Different preprocessing strategies show that human decisions significantly impact predictive performance. We show that the naive total sum scaling normalization that is commonly used in microbiome research is one of the optimal strategies to maximize predictive power. Also, we find that accurately defined labels are more important than normalization, taxonomic level, or model characteristics. ML performance is limited when humans can’t classify samples accurately. Lastly, we provide domain scientists via a full model selection decision tree to identify the human choices that optimize model prediction power. Conclusions Our study highlights the importance of incorporating diverse environmental features and careful data preprocessing in enhancing the predictive power of machine learning models for soil and biological phenotype connections. This approach can significantly contribute to advancing agricultural practices and soil health management.

Aghdam, Rosa↗
DOE OSTI2024

NANO.PTML model for read-across prediction of nanosystems in neurosciences. computational model and experimental case of study

Abstract Neurodegenerative diseases involve progressive neuronal death. Traditional treatments often struggle due to solubility, bioavailability, and crossing the Blood-Brain Barrier (BBB). Nanoparticles (NPs) in biomedical field are garnering growing attention as neurodegenerative disease drugs (NDDs) carrier to the central nervous system. Here, we introduced computational and experimental analysis. In the computational study, a specific IFPTML technique was used, which combined Information Fusion (IF) + Perturbation Theory (PT) + Machine Learning (ML) to select the most promising Nanoparticle Neuronal Disease Drug Delivery (N2D3) systems. For the application of IFPTML model in the nanoscience, NANO.PTML is used. IF-process was carried out between 4403 NDDs assays and 260 cytotoxicity NP assays conducting a dataset of 500,000 cases. The optimal IFPTML was the Decision Tree (DT) algorithm which shown satisfactory performance with specificity values of 96.4% and 96.2%, and sensitivity values of 79.3% and 75.7% in the training (375k/75%) and validation (125k/25%) set. Moreover, the DT model obtained Area Under Receiver Operating Characteristic (AUROC) scores of 0.97 and 0.96 in the training and validation series, highlighting its effectiveness in classification tasks. In the experimental part, two samples of NPs (Fe 3 O 4 _A and Fe 3 O 4 _B) were synthesized by thermal decomposition of an iron(III) oleate (FeOl) precursor and structurally characterized by different methods. Additionally, in order to make the as-synthesized hydrophobic NPs (Fe 3 O 4 _A and Fe 3 O 4 _B) soluble in water the amphiphilic CTAB (Cetyl Trimethyl Ammonium Bromide) molecule was employed. Therefore, to conduct a study with a wider range of NP system variants, an experimental illustrative simulation experiment was performed using the IFPTML-DT model. For this, a set of 500,000 prediction dataset was created. The outcome of this experiment highlighted certain NANO.PTML systems as promising candidates for further investigation. The NANO.PTML approach holds potential to accelerate experimental investigations and offer initial insights into various NP and NDDs compounds, serving as an efficient alternative to time-consuming trial-and-error procedures.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2026

Mitigating Algorithmic Bias in Cancer Site Classification Models

Purpose Integrating artificial intelligence in cancer diagnostics has improved tumor classification beyond rule-based systems. Despite these advancements, these models may still encode demographic biases. We conducted a large-scale, applied bias-probing study of a deep learning–based cancer site classifier to quantify race information encoded in document embeddings. We then evaluated how performance changes when race-correlated embedding dimensions are removed in a post-training sensitivity analysis. Methods The cancer site classifier was trained using 3.5 million electronic cancer pathology reports from six of the National Cancer Institute's SEER registries. We trained a hierarchical self-attention network to generate 400-dimensional document embeddings. These embeddings were used to train two downstream, gradient-boosted decision tree classifiers: one to classify the cancer sites and another to predict racial categories. We identified overlapping features by intersecting the top 50 feature-importance rankings from the site and race models and computed their cumulative feature importance in each model. As a post hoc sensitivity analysis, we progressively pruned these overlapping dimensions, retrained the site model, and compared overall macro-F1 and accuracy, race-stratified macro-F1, and group fairness metrics on the basis of demographic parity and equalized odds before and after pruning. Results The analysis revealed minimal feature overlap between the cancer site and race prediction models, and the cumulative importance scores indicated a negligible influence of racial information on clinical predictions. Post-training pruning of overlapping features did not compromise the models' diagnostic accuracy, with a 0.07% loss in accuracy. Conclusion Our findings demonstrate that HiSAN-generated embeddings from SEER data can be used effectively in cancer site classification without significant demographic bias influencing the outcomes. Post-training pruning therefore functions as a practical audit and sensitivity check.

Shivanna, Abhishek [ORNL] (ORCID:0009000665228593)↗