Simulated Data Research Articles

Reverse design of load-bearing broadband metamaterial absorber assisted by deep learning

Abstract In response to the current challenges of narrow absorption bandwidth, weak load-bearing capacity, and low design efficiency in absorbing structures, this study focuses on the reverse design of load-bearing broadband metamaterial absorber. A parameterized model of load-bearing metamaterial absorber was developed by integrating the composite sandwich structure with the electromagnetic resonant layers. The resonant layer was constructed using the combination of Vicsek-fractal and circular rings, with resistive films employed to broaden the absorption bandwidth. A deep learning-based forward prediction model was established to accurately predict the absorbance of the metamaterial absorber. The SHapley Additive exPlanations (SHAP) framework was utilized to analyze the forward prediction network, revealing the influence of various design parameters on the absorbance at center frequencies across the L to K band spectrum. Additionally, the Group Teaching Optimization Algorithm (GTOA) was introduced into the design process, leading to the development of an automated reverse design method for metamaterial absorber that can achieve specific design objectives. Using the GTOA-based reverse design method, a metamaterial absorber capable of effectively absorbing vertically incident electromagnetic waves within the 3-20 GHz frequency range was designed. The designed absorbing structure was fabricated, and its absorption performance was measured using the arch method. The measurement results were found to be in good agreement with the simulation data. The absorbing mechanism of the designed metamaterial absorber was analyzed based on the calculation of equivalent electromagnetic parameters and the electromagnetic resonance observed at the resonant frequency. It was determined that the primary absorbing effect is induced by electric resonance triggered by electromagnetic waves. The proposed metamaterial absorber can be applied to radar stealth design for military targets such as naval vessels. The research methodology and approach demonstrate significant generalizability and engineering applicability.

Causal discovery from nonstationary time series

AbstractThis paper introduces a new causal structure learning method for nonstationary time series data, a common data type found in fields such as finance, economics, healthcare, and environmental science. Our work builds upon the constraint-based causal discovery from nonstationary data algorithm (CD-NOD). We introduce a refined version (CD-NOTS) which is designed specifically to account for lagged dependencies in time series data. We compare the performance of different algorithmic choices, such as the type of conditional independence test and the significance level, to help select the best hyperparameters given various scenarios of sample size, problem dimensionality, and availability of computational resources. Using the results from the simulated data, we apply CD-NOTS to a broad range of real-world financial applications in order to identify causal connections among nonstationary time series data, thereby illustrating applications in factor-based investing, portfolio diversification, and comprehension of market dynamics.

Realtime estimation of average sand grain diameter and sand bed height and mass for a trailing suction hopper dredger.

A Trailing Suction Hopper Dredger (TSHD) in the sailing process, excavating seabed sediment with the drag head and pumping it through the pipeline to the spoil hopper sedimentation. The spoil hopper sedimentation process, as a critical factor affecting the loading yield and efficiency of a dredger, is susceptible to the influence of different average sand grain diameter . However, the soil type tends to change continuously during the loading process, making it difficult to obtain the by measurement or calculation. In this paper, combined with the existing simplified one-dimensional sedimentation model and simulated loading data, the Bootstrap Particle Filter (BPF), Gaussian Particle Swarm Optimized Particle Filter (GPSO-PF), Assisted Particle Filter (APF), Continuous-Discrete Feedback Particle Filter based on the constant feedback gain approximation (CD-FPF-CG) and the Galerkin method feedback gain approximation (CD-FPF-GL) are applied in the numerical simulations for online estimation of important variables such as the average sand grain diameter , the sand bed height and mass , respectively. The simulations results show that the CD-FPF-CG has a higher numerical accuracy and efficiency than the BPF, GPSO-PF, APF and CD-FPF-GL. Therefore, the CD-FPF-CG is used to estimate the above important variables for the actual vessel and the initial value of the average sand grain diameter for the CD-FPF-CG is determined via Simulated Annealing Multiple Population Genetic Algorithm (SAMPGA).

Maximum non-bounded difference method for overbounding Global Navigation Satellite System errors

AbstractGlobal Navigation Satellite Systems (GNSS) support numerous applications, including mission-critical ones that require high integrity for safe operations such as air, maritime, and land-based navigation. These applications require ensuring the upper bound of the positioning computation, including overbounding any assumptions made in positioning algorithms. One of the main assumptions in the GNSS positioning algorithm is that the error follows a Probability Density Function (PDF), widely assumed to be Gaussian. This assumption should be overbounded to ensure safety. Two mathematical definitions of distribution overbounding have been proposed: Cumulative Distribution Function (CDF) overbounding and paired overbounding. Achieving these definitions has been an open problem for the past 20 years, and to date, no methods have effectively achieved either definition in multimode datasets. This paper proposes the first overbounding method, the Maximum non-Bounded Difference (MnBD) method, capable of achieving both definitions. Its effectiveness is assessed by comparison with the combined CDF bounding and paired overbounding approach. The results show that the MnBD method successfully achieves overbounding in both real datasets and 1000 simulated multimode data, tested on Gaussian and Generalized Extreme Value (GEV) distributions. The latter demonstrates MnBD’s ability to handle asymmetric distributions. The developed approach can be utilized in overbounding the GNSS error at the system level and user level to ensure the system integrity.

An efficient deep learning method for amino acid substitution model selection.

Amino acid substitution models play an important role in studying the evolutionary relationships among species from protein sequences. The amino acid substitution model consists of a large number of parameters; therefore, it is estimated from hundreds or thousands of alignments. Both general models and clade-specific models have been estimated and widely used in phylogenetic analyses. The maximum likelihood method is normally used to select the best fit model for a specific protein alignment under the study. A number of studies have discussed theoretical concerns as well as computational burden of the maximum likelihood methods in model selection. Recently, machine learning methods have been proposed for selecting nucleotide models. In this paper, we propose a method to measure substitution rates among amino acids (called summary statistics) from protein alignments to efficiently train a deep learning network of so-called ModelDetector for detecting amino acid substitution models. The ModelDetector network was trained from 2,246,400 alignments on a computer with 8 cores (without GPU) in about 3.3 hours. Experiments on simulation data showed that the accuracy of the ModelDetector was comparable with that of the maximum likelihood method ModelFinder. It was orders of magnitude faster than the maximum likelihood method in inferring amino acid substitution models and able to analyze genome alignments with millions of sites in minutes. The results indicate that the deep learning network can play as a promising tool for amino acid substitution model selection.

Bioinformatics analysis to design a multi-epitope mRNA vaccine againstS. agalactiaeexploiting pathogenic proteins.

Antibiotic resistance in bacterial pathogen infections is a growing global issue that occurs due to their adaptation to changing environmental conditions. Therefore, producing an efficient vaccine as an alternative approach can improve the immune system, eradicate related pathogens, and overcome this growing problem.Streptococcus agalactiaebelongs to group B Streptococcus (GBS). Colonization of GBS during pregnancy is a significant risk factor for infants and young children. S. agalactiae infected population exhibits resistance to beta-lactams, including penicillin and the second-line antibiotics erythromycin and clindamycin.On the other hand, there are currently no commercial vaccines against this pathogen. Vaccination of pregnant women is a highly effective method to protect newborns and infants from S. agalactiaeinfection, and it has been identified as an urgent demand by the World Health Organization. This study employed various immunoinformatic tools to develop an effective vaccine that could trigger both humoral and cell-mediated immunity and prevent disease. For this purpose, three conserved antigenic proteins of the main pathogenic strains ofS. agalactiaewere utilized to predict CTL, HTL, and B-cell epitopes for producing an mRNA vaccine against different strains ofS. agalactiae. The selected epitopes were fused using proper linkers. The Resuscitation promoting factor E (RpfE) sequence was incorporated in the designed vaccine construct as an adjuvant to boost its immune response. Different physicochemical characteristics of the final designed vaccine, modeling of the three-dimensional structure, molecular docking, molecular dynamics simulation, and immunological response simulation were screened following vaccine administration in an in vivo model.Computational immune simulation data identified that IgG1, IgM, INF γ, IL-2, T helper, and B-cell populations increased significantly after vaccination. These findings suggested that the vaccine candidate may provide good protection againstS. agalactiaeinfection. However,experimental and animal model studies are required for additional validation and implementation in human vaccination programs.

Assessing the impact of hard data patterns on Bayesian Maximum Entropy: a simulation study.

This study empirically tested the robustness of Bayesian Maximum Entropy (BME) in predicting spatiotemporal data, with an emphasis on skewness, sample size, and spatial dependency level. Simulated data, both Gaussian and non-Gaussian, were generated using the unconditional sequential simulation method, with sample sizes ranging from 100 to 500 at the interval length of 50 and varying skewness (0, 1, 3, 6 and 9) and spatial dependency levels (weak, moderate, and strong). Findings revealed sample size variations and spatial dependence levels did not significantly influence BME prediction's Mean Square Error (MSE) and bias. While skewness significantly impacted MSE (p-value < 0.001), bias remained unaffected. Moreover, skewness and spatial dependence interactions affected both MSE and bias. Despite this, BME proved robust to sample size and skewness, demonstrating a negligible MSE on the graphical plot (heatmap).

Refraction-Aware Structure from Motion for Airborne Bathymetry

In this work, we introduce the first pipeline that combines a refraction-aware structure from motion (SfM) method with a deep learning model specifically designed for airborne bathymetry. We accurately estimate the 3D positions of the submerged points by integrating refraction geometry within the SfM optimization problem. This way, no refraction correction as post-processing is required. Experiments with simulated data that approach real-world capturing conditions demonstrate that SfM with refraction correction is extremely accurate, with submillimeter errors. We integrate our refraction-aware SfM within a deep learning framework that also takes into account radiometrical information, developing a combined spectral and geometry-based approach, with further improvements in accuracy and robustness to different seafloor types, both textured and textureless. We conducted experiments with real-world data at two locations in the southern Mediterranean Sea, with varying seafloor types, which demonstrate the benefits of refraction correction for the deep learning framework. We made our refraction-aware SfM open source, providing researchers in airborne bathymetry with a practical tool to apply SfM in shallow water areas.

Optimization of working parameters of high-pressure roller mill based on entropy weight method and response surface method.

In order to improve the crushing efficiency of high-pressure roller mill and reduce energy consumption, the optimal parameter combination of high-pressure roller mill is sought, GM160-140 high-pressure roller mill is taken as the research object, and the discrete element simulation calculation model of high-pressure roller mill is established to simulate the crushing process. Taking the roll surface pressure, productivity, and the proportion of discharge particle diameter as the performance evaluation indexes, the simulation data are used to analyze the influence of each working condition parameter of the high-pressure roller mill on its performance law. Through the entropy weighting method to determine the weights of high-pressure roller mill productivity, roll surface pressure, the proportion of discharge particle diameter, and through the extreme difference analysis to get the optimal parameter combination. Minitab software is used to carry out the response surface method (RSM) experimental design, the optimal parameter combinations are derived from the optimization using the response surface apparatus, the optimal parameter combinations derived from the two optimization methods are compared and analyzed, and ultimately it is concluded that the optimization of the response surface is the global optimal, and the optimal parameter combinations are as follows: the roller speed is 2.50rad/s, the roller gap is 28.87mm, and the feed size is 31.97mm, Under the above optimal parameters, the roll surface pressure is 150450.00N, the productivity is 1268.00t/h, the percentage of the material less than 6mm is 58.14%, so as to achieve the purpose of optimizing the parameters of the high-pressure roller mill, It is of significance to save resources and respond to the national green development and intelligent development.

Modeling and Estimation of Adaptive Optics-Corrected long-exposure Point Spread Function Using Gaussian Sum Approximation

Abstract Adaptive optics (AO) enhances astronomical images by correcting perturbations in the optical system, resulting in complex point spread function (PSF) shapes, which provide critical information for optical design, calibration, and diagnostics. This paper presents a parametric model for an AO-corrected long-exposure PSF that adapts to complex shapes and various seeing conditions. A complementary estimation method is also introduced, addressing the inverse problem of estimating the atmospheric PSF or power spectral density (PSD) from observed PSF data. The model, based on Gaussian sum approximation (GSA), was tested using simulated PSF data from the OOMAO toolbox for Fried parameter (r0) values ranging from 0.10 to 0.20 m and on-sky data from the Very Large Telescope’s MUSE instrument. Two approaches were analyzed: (i) direct PSF estimation using GSA, and (ii) PSF estimation derived from atmospheric PSD estimation. Both methods were evaluated using root mean square error (RMSE). While approach (i) provided the best performance, approach (ii) produced more accurate atmospheric PSF estimates compared to the Moffat approximation and the Fetick method. Additionally, approach (ii) enabled the estimation of key physical parameters, such as the Fried parameter (r0) and residual AO variance ($\sigma _{AO}^2$). The proposed methods effectively estimate atmospheric PSF and PSD, even for complex PSF shapes influenced by spiders, central obstructions, or static aberrations.

AncestryPainter 2.0: Visualizing Ancestry Composition and Admixture History Graph.

The earlier vertion of AncestryPainter is a Perl program to display the ancestry composition of numerous individuals using a rounded graph. Motivated by the requests of users in practical applications, we updated AncestryPainter to version 2.0 by coding in an R package and improving the layout, providing more options and compatible statistical functions for graphing. Apart from improving visualization functions per se in this update, we added an extra graphing module to visualize genetic distance through radial bars of varying lengths surrounding a core. Notably, AncestryPainter 2.0 implements a method admixture history graph (AHG) to infer the admixture sequence of multiple ancestry populations, and allows for multiple pie charts at the center of the graph to display the ancestry composition of more than one target population. We validated the six AHG metrics using both simulated and real data and implemented a Pearson coefficient-based metric with the best performance in AncestryPainter 2.0. Furthermore, a statistical module to merge ancestry proportion matrices. AncestryPainter 2.0 is freely available at https://github.com/Shuhua-Group/AncestryPainterV2 and https://pog.fudan.edu.cn/#/Software.

Addressing statistical challenges in the analysis of proteomics data with extremely small sample size: a simulation study.

One of the most promising approaches for early and more precise disease prediction and diagnosis is through the inclusion of proteomics data augmented with clinical data. Clinical proteomics data is often characterized by its high dimensionality and extremely limited sample size, posing a significant challenge when employing machine learning techniques for extracting only the most relevant information. Although there is a wide array of statistical techniques and numerous analysis pipelines employed in proteomics data analysis, it is unclear which of these methods produce the most efficient, reproducible, and clinically meaningful results. In this study, we compared 9 unique analysis schemes comprised of different machine learning and dimensionality reduction methods for the analysis of simulated proteomics data consisting of 1317 proteins measured in 26 subjects (i.e., 13 controls and 13 cases). In scenarios where the sample size is extremely small (i.e., n < 30), all schemes resulted in an exceptionally high level of performance metrics, indicating potential overfitting. While performance metrics did not exhibit significant differences across schemes, the set of proteins selected to be discriminatory between groups demonstrated a substantial level of heterogeneity. However, despite heterogeneity in the selected proteins, their biological pathways and genetic diseases exhibited similarities. A sensitivity analysis conducted using varying sample sizes indicated that the stability of a set of selected biomarkers improves with larger sample sizes within a scheme. When the aim of the study is to identify a statistical model that best distinguishes between cohort groups using proteomics data and to uncover the biological pathways and disorders common among the selected proteins, the majority of widely used analysis pipelines perform similarly. However, if the main objective is to pinpoint a set of selected proteins that wield significant influence in discriminating cohort groups and utilize them for subsequent investigations, meticulous consideration is necessary when opting for statistical models, due to the possibility of heterogeneity in the sets of selected proteins.

Multi-condition hawser tension prediction of offshore offloading system based on long and short-term memory network and transfer learning

Real-time prediction of hawser tension of the side-by-side oil offloading system of Floating Production Storage and Offloading (FPSO) can provide early warnings for hawser breakages and ship collision risk, thus improving structural, property, and environmental security. Although offline numerical models and Long Short-Term Memory (LSTM) could offer substantial precision, they confront intensive time costs in recalibrating or retraining models. This paper proposes an integration method of LSTM networks and transfer learning for real-time tension prediction considering inputs of actual remote wave elevations. We use short-term environmental data and numerical simulation data of correlated hawser tensions during offloading operations to train a pre-trained benchmark model for transfer learning. Then, a highly generalized and efficient transferred model is constructed by using a small sample to realize short-term tension predictions in time-varying environments. The results show the occurrence time and value of extreme tensions predicted by transfer learning nearly match the reference data, and their maximum errors are 3 s and 0.11, respectively, superior to LSTM direct training. Therefore, it provides sufficient demand for real-time prediction and early collision risk prevention in dynamically changing ocean environment. The research results could provide an alternative framework for intelligent monitoring of large-scale marine structures.

Aircraft Parameter Estimation Considering Process and Measurement Noise

A practical formulation is proposed for aircraft parameter estimation using the filter-error method, which is a maximum likelihood estimator for dynamic systems with both process noise and measurement noise. The novelty of the proposed formulation is that by accurately estimating the measurement noise covariance matrix using a time series analysis method, the remaining unknowns (which include the unknown parameters in the state-space matrices and the process noise covariance matrix) become decorrelated and can be estimated simultaneously in a straightforward manner. The approach is demonstrated using simulation data and flight test data from a subscale airplane. Results indicate that the proposed algorithm can produce accurate modeling results when both measurement noise and process noise are present in the data.

Rapid parameter estimation for merging massive black hole binaries using continuous normalizing flows

Abstract Detecting the coalescences of massive black hole binaries (MBHBs) is one of the primary targets for space-based gravitational wave observatories such as laser interferometer space antenna, Taiji, and Tianqin. The fast and accurate parameter estimation of merging MBHBs is of great significance for the global fitting of all resolvable sources, as well as the astrophysical interpretation of gravitational wave signals. However, such analyses usually entail significant computational costs. To address these challenges, inspired by the latest progress in generative models, we explore the application of continuous normalizing flows (CNFs) on the parameter estimation of MBHBs. Specifically, we employ linear interpolation and trig interpolation methods to construct transport paths for training CNFs. Additionally, we creatively introduce a parameter transformation method based on the symmetry in the detector’s response function. This transformation is integrated within CNFs, allowing us to train the model using a simplified dataset, and then perform parameter estimation on more general data, hence also acting as a crucial factor in improving the training speed. In conclusion, for the first time, within a comprehensive and reasonable parameter range, we have achieved a complete and unbiased 11-dimensional rapid inference for MBHBs in the presence of astrophysical confusion noise using CNFs. In the experiments based on simulated data, our model produces posterior distributions comparable to those obtained by nested sampling.

Components and anisotropy of 3D QFP waves during the early solar eruption

Abstract The propagation of disturbances in the solar atmosphere is inherently three-dimensional (3D), yet comprehensive studies on the spatial structure and dynamics of 3D wavefronts are scarce. Here we conduct high-resolution 3D numerical simulations to investigate filament eruptions, focusing particularly on the 3D structure and genesis of EUV waves. Our results demonstrate that the EUV wavefront forms a dome-like configuration subdivided into three distinct zones. The foremost zone, preceding the flux rope, consists of fast-mode shock waves that heat the adjacent plasma. Adjacent to either side of the flux rope, the second zone contains expansion waves that cool the nearby plasma. The third zone, at the juncture of the first two, exhibits minimal disturbances. This anisotropic structure of the wavefront stems from the configuration and dynamics of the flux rope, which acts as a 3D piston during eruptions—compressing the plasma ahead to generate fast-mode shocks and evacuating the plasma behind to induce expansion waves. This dynamic results in the observed anisotropic wavefront. Additionally, with synthetic EUV images from simulation data, the EUV waves are observable in Atmospheric Imaging Assembly 193 and 211 $\AA$, which are identified as the fast-mode shocks. The detection of EUV waves varies with the observational perspective: the face-on view reveals EUV waves from the lower to the higher corona, whereas an edge-on view uncovers these waves only in the higher corona.

Molecular dynamics simulation of extraction of Curcuma longa L. extract using subcritical water.

Humans have utilized plants for various purposes, including sustenance and medical treatment for millennia. Researchers have extensively investigated medicinal plants' potential in drug development, spurred by their rich array of chemical compounds. Curcumin, a valuable bioactive compound, is extracted from Turmeric, known by the scientific name Curcuma Longa L. Notably, curcumin boasts potent antioxidant and anti-inflammatory properties, making it a promising candidate for treating cancer and other microbial diseases. Therefore, the simulation study of the extraction of this important medicinal compound by water, which is a green solvent, was carried out. This study employed molecular dynamics simulation for the first time to explore the extraction of Curcuma Longa L. extract using subcritical water. The simulations were carried out at constant pressure and different temperatures, using the Compass force field in the Lammps simulation package. The findings revealed an increase in the amount of Curcuma longa extract with rising temperature, indicating a weakening of hydrogen bonds in water molecules. Water lost its polar state with increasing temperature and became a suitable non-polar solvent for extracting non-polar compounds. The average absolute relative deviation (AARD) for calculated and simulated density data was 6.45%.

Dual-encoder architecture for metal artifact reduction for kV-cone-beam CT images in head and neck cancer radiotherapy.

During a radiotherapy(RT) course, geometrical variations of target volumes, organs at risk, weight changes (loss/gain), tumor regression and/or progression can significantly affect the treatment outcome. Adaptive RT has become the effective methods along with technical advancements in imaging modalities including cone-beam computed tomography (CBCT). Planning CT (pCT) can be modified via deformable image registration (DIR), which is applied to the pair of pCT and CBCT. However, the artifact existed in both pCT and CBCT is a vulnerable factor in DIR. The dose calculation on CBCT is also suggested. Missing information due to the artifacts hinders the accurate dose calculation on CBCT. In this study, we aim to develop a deep learning-based metal artifact reduction(MAR) model to reduce the metal artifacts in CBCT for head and neck cancer RT. To train the proposed MAR model, we synthesized the kV-CBCT images including metallic implants, with and without metal artifacts (simulated image data pairs) through sinogram image handling process. We propose the deep learning architecture which focuses on both artifact removal and reconstruction of anatomic structure using a dual-encoder architecture. We designed four single-encoder models and three dual-encoder models based on UNet (for an artifact removal) and FusionNet (for a tissue restoration). Each single-encoder model contains either UNet or FusionNet, while the dual-encoder models have both UNet and FusionNet architectures. In the dual-encoder models, we implemented different feature fusion methods, including simple addition, spatial attention, and spatial/channel wise attention. Among the models, a dual-encoder model with spatial/channel wise attention showed the highest scores in terms of peak signal-to-noise ratio, mean squared error, structural similarity index, and Pearson correlation coefficient. CBCT images from 34 head and neck cancer patients were used to test the developed models. The dual-encoder model with spatial/channel wise attention showed the best results in terms of artifact index. By using the proposed model to CBCT, one can achieve more accurate synthetic pCT for head and neck patients as well as better tissue recognition and structure delineation for CBCT image itself.

Enhancement of mannosylerythritol lipid-A on physicochemical stability, antioxidant activity, and bioavailability of bovine lactoferrin emulsion under different pH conditions.

This study systematically explored the enhancement of mannosylerythritol lipid-A (MEL-A) on physicochemical stability, antioxidant activity, and bioavailability of bovine lactoferrin (BLF) emulsion under different pH conditions by spectroscopic techniques, molecular simulation, and simulated in vitro digestion model. The bovine lactoferrin-MEL-A (BLF-MEL-A) emulsions were prepared and characterized with the Fourier infrared, of which results showed that high concentration MEL-A (1.00 mg/mL) changed the secondary structure of pH-induced BLF and rendered an increase in β-sheet and random coil fractions. Based on the results of fluorescence spectrum and isothermal titration calorimetry, hydrogen bonding, van der Waals forces, and electrostatic force were the interaction forces of BLF and MEL-A, which were similar to the simulated data of molecular docking and molecular dynamics. BLF-MEL-A emulsion also exerted considerable antioxidant activities and had great potential for functional food development. In addition, MEL-A could improve the stability of BLF emulsion in simulated in vitro digestion test, which promoted the bioavailability of BLF. Therefore, this study will facilitate to explore the interaction between BLF and MEL-A and expand the application of MEL-A as a food additive in food industry.

Abstract B065: FabricaTM: A large-scale data simulation platform isolates tumor signal from cell-free DNA and improves tissue of origin prediction accuracy

Abstract Cell-free DNA (cfDNA) liquid biopsy is a promising non-invasive method for disease detection, but data availability and the complexity of cfDNA composition pose barriers to model development. Cancer prediction models trained on cfDNA data from clinically collected blood samples often struggle to isolate tumor-derived signals from confounding factors, and tissue of origin (TOO) models suffer from small sample sizes of rarer tumor types. To address these challenges, we developed FabricaTM, a platform to generate diverse, large-scale, high-fidelity simulated cfDNA datasets for robust and accurate machine learning (ML) model training. A key advantage of FabricaTM is its ability to enhance ML models by matching confounding variables between non-cancer and cancer samples. We first synthesized age-balanced non-cancer samples, then simulated tumor DNA shedding into circulation from 502 reference biopsy samples across 16 indications, optimizing for final tumor content distributions ranging from 0.001 to 14.2%. All reference non-cancer and biopsy samples consisted of aligned bisulfite sequencing reads from a custom target hybrid capture panel. This approach expanded a non-cancer dataset of 177 samples to 1,212 simulated samples and generated 6,400 simulated cancer cfDNA samples matching the non-cancers in age distribution. We then assessed this dataset for equivalence with clinical data and ability to improve cancer prediction. Non-linear dimensionality reduction and clustering of methylation signal showed simulated samples cluster indistinguishably from 2,290 samples (1,149 non-cancer, 1,141 treatment-naïve cancer) from the CORE-HH clinical study (NCT05435066). A binary cancer prediction model trained on the FabricaTM-generated dataset and evaluated on the CORE-HH dataset showed reduced reliance on age-associated signal (R2=0.15) and increased tuning towards tumor content, with minor cost to sensitivity (&amp;lt;4% at 95% specificity), relative to a model trained and cross-validated on the CORE-HH data (R2=0.60). To evaluate the benefit of these data for TOO prediction, we trained multiclass TOO models using FabricaTM's tumor biopsy reference data with and without the addition of our simulated data for 10 target indication groupings and benchmarked their performance on our clinical dataset. Training with both data types yielded a 10% increase in balanced accuracy. Moreover, peak performance was achieved at different training sample numbers per indication, illustrating FabricaTM's potential to estimate sample size requirements during study design. Our findings suggest FabricaTM-generated data can augment limited datasets to train ML models to identify tumor-specific biomarkers obscured by technical and demographic biases in real-world data. The age-balancing approach is readily applied to other confounders to help models learn true disease signatures. The platform is also adaptable to other diseases and biofluids (e.g., Alzheimer's disease, urine), allowing for extension to diagnostic and screening development beyond cancer and blood across the care spectrum. Citation Format: Kade Pettie, Shiva Farashahi, Jackson Killian, Dorna Kashef, Josh Hubbell, Feras Hantash, Jocelyn Charlton, Kieran Chacko. FabricaTM: A large-scale data simulation platform isolates tumor signal from cell-free DNA and improves tissue of origin prediction accuracy [abstract]. In: Proceedings of the AACR Special Conference: Liquid Biopsy: From Discovery to Clinical Implementation; 2024 Nov 13-16; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(21_Suppl):Abstract nr B065.