Data Model Research Articles

The impact of perceived experience with ChatGPT on online consumers' information searching behavior: an empirical study of Iranian college students

PurposeThe purpose of this study is to explore the impact of perceived experience with ChatGPT on online consumers' information searching behavior. The study also examines the moderating effects of those relationships.Design/methodology/approachPrimary data were collected through an online survey. In total, 370 eligible responses were received. This study applied partial least squares structural equation modeling (PLS-SEM) for data analysis.FindingsThe findings demonstrate that both perceived personalization and perceived relevance have a direct impact on online consumers' information searching behaviors. Additionally, the results also indicate that perceived accuracy and perceived convenience lead to positive online consumers' information searching behavior. Moreover, age and gender and education level play moderating mechanisms in this relationship.Practical implicationsOverall, the conclusions of this study provide valuable insights into the potential of ChatGPT to improve online consumers behavior. However, there are still many unanswered questions about the impact of ChatGPT on consumers' experience. Future research is needed to explore these questions and to further our understanding of the potential of ChatGPT to revolutionize online information searching and digital marketing.Originality/valueTo the best of the authors’ knowledge, this paper is the first of its kind to highlight the impact of perceived experience with ChatGPT on online consumers' information searching behavior of its extensive use in scientific research and academic work. The importance of this study lies in the fact that it presents the behaviors concerns and future fears of people in academia as they cope with and deal with the inevitable reality of artificial intelligence (AI) language models such as ChatGPT.

X-ray view of dissipative warm corona in active galactic nuclei

Context. Detections of the X-ray spectra of active galactic nuclei (AGNs) typically reveal a noticeable excess of soft X-rays beyond the extrapolation of the power-law trend observed between 2 and 10 keV. However, the cause of this surplus remains unclear. In the scenario of soft Comptonization, observations suggest a warm corona temperature, ranging from 0.1 to 1 keV, and an optical depth of approximately 10 to 30. Furthermore, according to radiative constraints derived from spectral analyses employing Comptonization models, it has been suggested that most of the accretion power is released within the warm corona. At the same time, the disk beneath it is largely non-dissipative, mainly emitting the reprocessed radiation from the corona. Aims. We tested the dissipative warm corona model using the radiative transfer code TITAN-NOAR on a sample of 82 XMM-Newton EPIC-pn observations of 21 AGNs. Based on a spectral modeling of the X-ray data, we aim to estimate the total amount of internal heating inside the warm corona on top of the accretion disk. Methods. By modeling the 0.3–10 keV EPIC-pn spectra with the TITAN-NOAR model component, we estimated the internal heating and optical depth of the warm corona and checked their correlations with such global parameters as the hot corona spectral index, black hole mass, and accretion rate. From the model normalization, we computed the radial extent of the warm corona on top of the cold accretion disk. Results. Our model infers the presence of dissipative warm corona, with optical depths distributed across the range of ∼6–30 and a total internal heating in the range of ∼1–29 × 10−23 erg s−1 cm3. We do not detect any variation between these properties and global properties, such as the black hole mass and accretion rate. The extent of the warm corona is spread across a broad range, from 7–408 gravitational radii, and we find that the warm corona is more extended for higher accretion rates. Conclusions. Soft excess emission is ubiquitous across a wide mass range and accretion rate in AGNs. We confirm that the warm corona responsible for producing the soft X-ray excess is highly dissipative with greater optical depths being associated with lower internal heating and vice versa. The cold standard accretion disk regulates the extent of the warm corona.

3D stochastic simulation of a deep-water turbidite system: an example from the Los Molles Formation, southern region of the Neuquén Basin, Argentina

A recurring challenge in geological modeling is bridging the gap between different scales. For example, difficulties arise when connecting outcrop data through production to exploration scale models. An object-based stochastic simulation was performed using outcrop data from the Arroyo La Jardinera area in the southern region of the Neuquén Basin, Argentina. This simulation involved four depositional sequences in a vertical succession that includes turbidites and associated deep marine facies. The aims of this study are (a) to determine the best geological model consistent with field and aerial image data; (b) to validate the application of object modeling to determine facies distribution; (c) to evaluate uncertainties from models based on scarce data. The studied interval covers a transgressive (J1) to regressive succession (J21, J22, and J23) of basin plain to slope depositional settings, featuring sandy and gravelly turbidite channels, turbidite lobes and interlobes, lobe fringes, and muddy slope and basin plain. Each depositional sequence model was constructed using specific input parameters for architectural elements, with lithological proportions based on sedimentary logs. The J1 sequence includes basin plain, lobe fringe, and minor lobe deposits; J21 features turbidite lobes, fringes, and subordinate channels; J22 accommodates turbidite channels scoured into muddy slope facies; and J23 encompasses gravelly and sandy turbidite channels carved on muddy slope facies. The geostatistical modeling of outcrop data has allowed building a quantitative sedimentological model useful for understanding subsurface facies heterogeneity in both exploration (vertical) and production (horizontal) scales.

Updates to the wild bird abundance and movement models for the early warning system for avian influenza in the EU

Abstract Highly pathogenic avian influenza (HPAI) viruses pose a significant threat to both poultry and wild bird populations. Migratory wild birds play a key role in the intercontinental spread of avian influenza (AI), introducing the virus into poultry populations. In response to frequent AI outbreaks in Europe, the European Food Safety Authority (EFSA), at the request of the European Commission (EC), produces quarterly and annual epidemiological reports to monitor and analyse AI trends. A key component of this surveillance includes the integration of outbreak data from Member States and contributions from non‐governmental ornithological organisations like the European Bird Census Council (EBCC) and the European Union for Bird Ringing (EURING) together in a predictive spatio‐temporal risk assessment model. Previous data integration and modelling efforts led to the development of an early warning system for predicting HPAI outbreaks accessible through a publicly available online user interface: the Bird Flu Radar. This report presents an improvement of the system by expanding the species coverage and refining the existing base models behind the epidemiological model. Specifically, this report details the exploration to incorporate 12 additional wild bird species into the models, and the changes made to the base models predicting the distribution and movements of wild birds. We demonstrate the improvements respecting the existing base models while at the same time enhancing the effectiveness in predicting HPAI outbreaks and possibly mitigating negative effects in Europe by providing more accurate predictions to different stakeholders.

Device-Performance-Driven Heterogeneous Multiparty Learning for Arbitrary Images.

Multiparty learning (MPL) is an emerging framework for privacy-preserving collaborative learning. It enables individual devices to build a knowledge-shared model and remaining sensitive data locally. However, with the continuous increase of users, the heterogeneity gap between data and equipment becomes wider, which leads to the problem of model heterogeneous. In this article, we concentrate on two practical issues: data heterogeneous problem and model heterogeneous problem, and propose a novel personal MPL method named device-performance-driven heterogeneous MPL (HMPL). First, facing the data heterogeneous problem, we focus on the problem of various devices holding arbitrary data sizes. We introduce a heterogeneous feature-map integration method to adaptively unify the various feature maps. Meanwhile, to handle the model heterogeneous problem, as it is essential to customize models for adapting to the various computing performances, we propose a layer-wise model generation and aggregation strategy. The method can generate customized models based on the device's performance. In the aggregation process, the shared model parameters are updated through the rules that the network layers with the same semantics are aggregated with each other. Extensive experiments are conducted on four popular datasets, and the result demonstrates that our proposed framework outperforms the state of the art (SOTA).

240 (PB228): Data analysis and model evaluation of chemotherapy induced neutropenia – results from the PARTNER trial

240 (PB228): Data analysis and model evaluation of chemotherapy induced neutropenia – results from the PARTNER trial

Finance Modeling Approach Using Machine Learning

The integration of machine learning (ML) into financial modeling represents a transformative shift in the finance sector, enhancing analytical capabilities and decision-making processes. This article explores the fundamental concepts of financial modeling and machine learning, illustrating how ML methodologies—both supervised and unsupervised—improve traditional financial models by enabling the analysis of extensive datasets to uncover intricate patterns. By leveraging historical data, ML enhances predictive accuracy in various applications, including risk assessment, algorithmic trading, credit scoring, and fraud detection. The article highlights significant advancements in predictive analytics through techniques such as regression analysis and time series forecasting, which allow financial analysts to navigate market uncertainties more effectively. Moreover, the automation of data processing and feature engineering through ML leads to increased operational efficiency, reducing human error and enhancing real-time decision-making capabilities. However, challenges such as data quality issues, model interpretability, and ethical considerations regarding algorithmic bias necessitate careful management to ensure responsible implementation. The article emphasizes the importance of transparency in ML models to foster trust among stakeholders while meeting regulatory requirements. Future research directions are proposed, focusing on the integration of explainable AI with blockchain technology and the implications of big data analytics on ML models. By addressing these emerging challenges and opportunities, finance professionals can leverage advanced analytical tools to drive innovation and enhance strategic initiatives in an increasingly complex financial landscape. Ultimately, this article underscores the pivotal role of machine learning in reshaping financial modeling practices and its potential to redefine the future of finance.

Unraveling the impact of environmentally sustainable practices of exhibitions and green behavior of visitors：The role of green psychological climate and environmental felt responsibility

To minimize environmental harm and enhance their competitive advantage, a growing number of exhibitions are adopting environmentally sustainable practices. While there is a growing body of research on exhibitions adopting environmentally sustainable practices, visitors’ perceptions of and responses to these practices remain complex and under-explored. This study explores how visitors' perceptions of environmentally sustainable practices at exhibitions contribute to green behavior. The mediating mechanism of green psychological climate and environmental felt-responsibility and the moderation of visitation frequency in this relationship are also discussed. The proposed hypothesis was tested using structural equation modeling of data collected from visitors. The results of the survey (N=531) indicate that an exhibition utilizing environmentally sustainable practices has a positive, direct effect on green behavior of its visitors, as well as indirect effects through their green psychological climate and environmental felt responsibility. Moreover, when experienced visitors perceive environmentally sustainable practices, this has a greater impact on their green psychological climate and environmental felt responsibility. With these insights reinforced by a multi-group analysis, they can guide exhibition organizers in implementing more effective environmentally sustainable practices to evoke environmental felt responsibility in visitors and encourage green behavior.

Surface insight: Leveraging high-density dataset fusion for enhanced roughness classification

The ability to assess the surface quality quickly and accurately is of immense importance in manufacturing system. Modern metrology system along with machine learning is great at classification but requires more time. Traditionally accessing surface roughness is a time-consuming process. The progress in manufacturing technology necessitates improved approaches for quality control, specifically in the categorization of surface roughness, which has a substantial impact on the performance of materials. This research study introduces a novel method for classifying surface roughness by combining image data and point cloud data to create a comprehensive model. It then compares the performance of this model with a model that just relies on image data. A comprehensive analysis is conducted in this study, where image and point cloud data is collected and analysed. Multilinear principal component analysis (MPCA) along with random forest classifier is employed to create a model that classifies the surface texture. The primary goal is to showcase the enhanced precision and comprehensive understanding offered by the fused data model compared to the model that solely relies on images.Furthermore, the work presents a pragmatic approach for developing this enhanced model offline and applying it online in real-time production environments, with a particular focus on using only image data. This strategy is in line with the objectives of Industry 4.0, which seeks to achieve more intelligent and data-driven manufacturing processes. Subsequent investigations will prioritize expanding the model’s suitability to various manufacturing settings, particularly highlighting its capacity to ensure quality in manufacturing lines through the utilization of images.

Co-creation and evaluation of an algorithm for the development of a mobile application for wound care among new graduate nurses: A mixed methods study.

Chronic wounds are a growing concern due to aging populations, sedentary lifestyles and increasing rates of obesity and chronic diseases. The impact of such wounds is felt worldwide, posing a considerable clinical, environmental and socioeconomic challenge and impacting the quality of life. The increasing complexity of care requires a holistic approach, along with extensive knowledge and skills. The challenge experienced by health-care professionals is particularly significant for newly graduate nurses, who face a gap between theory and practice. Digital tools, such as mobile applications, can support wound care by facilitating more precise assessments, early treatment, complication prevention and better outcomes. They also aid in clinical decision-making and improve healthcare delivery in remote areas. Several mobile applications have emerged to enhance wound care. However, there are no applications dedicated to newly graduate nurses. The aim of this study was to co-create and evaluate an algorithm for the development of a wound care mobile application supporting clinical decisions for new graduate nurses. The development of this mobile application is envisioned to improve knowledge application and facilitate evidence-based practice. This study is part of a multiphase project that adopted a pragmatic epistemological approach, using the 'Knowledge-to-Action' conceptual model and Duchscher's Stages of Transition Theory. Following a scoping review, an expert consensus, and stakeholder meetings, this study was pursued through a sequential exploratory mixed methods design carried out in two phases. In the initial phase, 21 participants engaged in semi-structured focus groups to explore their needs regarding clinical decision support in wound care, explore their perceptions of the future mobile application's content and identify and categorize essential components. Through descriptive analysis, five overarching themes emerged, serving as guiding principles for conceptual data model development and refinement. These findings confirmed the significance of integrating a comprehensive glossary complemented by photos, ensuring compatibility between the mobile application and existing documentation systems, and providing quick access to information to avoid burdening work routines. Subsequently, the algorithm was created from the qualitative data collected. The second phase involved presenting an online SurveyMonkey® questionnaire to 34 participants who were not part of the initial phase to quantitatively measure the usability of this algorithm among future users. This phase revealed very positive feedback regarding the usability [score of 6.33 (±0.19) on a scale of 1-7], which reinforces its quality. The technology maturation process can now continue with the development of a prototype and subsequent validation in a laboratory setting.

Advances and prospects of multi-modal ophthalmic artificial intelligence based on deep learning: a review.

In recent years, ophthalmology has emerged as a new frontier in medical artificial intelligence (AI) with multi-modal AI in ophthalmology garnering significant attention across interdisciplinary research. This integration of various types and data models holds paramount importance as it enables the provision of detailed and precise information for diagnosing eye and vision diseases. By leveraging multi-modal ophthalmology AI techniques, clinicians can enhance the accuracy and efficiency of diagnoses, and thus reduce the risks associated with misdiagnosis and oversight while also enabling more precise management of eye and vision health. However, the widespread adoption of multi-modal ophthalmology poses significant challenges. In this review, we first summarize comprehensively the concept of modalities in the field of ophthalmology, the forms of fusion between modalities, and the progress of multi-modal ophthalmic AI technology. Finally, we discuss the challenges of current multi-modal AI technology applications in ophthalmology and future feasible research directions. In the field of ophthalmic AI, evidence suggests that when utilizing multi-modal data, deep learning-based multi-modal AI technology exhibits excellent diagnostic efficacy in assisting the diagnosis of various ophthalmic diseases. Particularly, in the current era marked by the proliferation of large-scale models, multi-modal techniques represent the most promising and advantageous solution for addressing the diagnosis of various ophthalmic diseases from a comprehensive perspective. However, it must be acknowledged that there are still numerous challenges associated with the application of multi-modal techniques in ophthalmic AI before they can be effectively employed in the clinical setting.

Real-world prototyping and evaluation of digital twins for predictive facility maintenance

Digital Twins (DTs) hold promise for the effective implementation of predictive maintenance (PdM) in constructed facilities. However, their adoption in the AECO industry is lagging due to the low level of knowledge on their design and implementation, influenced by the lack of real-world exemplars. This paper presents two real-world BIM-based DTs for PdM in an existing building, detailing the methodology for their design and implementation. The two DT designs cover sensor data integration from legacy systems with digital models of the facility and the programming of analytics for PdM as well as data modeling for representing the physical objects in the digital model. In addition, the paper evaluates the two DT solutions and provides key considerations to make when deciding whether to build or buy. Overall, the study makes contributions towards criteria for DT platform selection, “build versus buy” comparisons, system architecture design, and end-user feedback on DT user interactions and task-dependent performance.

Magnetic recyclable carboxymethyl cellulose/gelatin/citrate@Fe3O4 photo-nanocomposite beads for ciprofloxacin removal via hybrid adsorption/photocatalysis process under solar light as a renewable energy source

Magnetically separable cross-linked carboxymethyl cellulose/gelatin/citrate-functionalized magnetite nanoparticles (Cit-Fe3O4) photo-nanocomposite beads (mCMC/Ge) were synthesized and applied in synergistic adsorption/photocatalytic degradation of ciprofloxacin (Cipro) pharmaceutical pollutant under sunlight irradiation. Various analytical techniques were employed to characterize their structural, textural, magnetic, thermal, and optical properties. The removal efficiency of mCMC/Ge beads was investigated considering different influencing parameters (pH, beads dosage, contact time, Cipro concentration, and temperature). Experimental data modeling indicated that the adsorption process followed pseudo-second-order kinetics and Langmuir isotherm models, with a maximum Langmuir adsorption capacity (qm) of 50 mg g−1 for mCMC/Ge, twice that of the matrix. Photocatalytic activity results showed prominent enhancement in Cipro removal using 1 g L−1 of mCMC/Ge at pH 7, as compared to Cit-Fe3O4, reaching 96 %, 85 %, and 63 % after 180 min of adsorption and 120 min of irradiation for initial pollutant concentrations of 10, 20, and 60 mg L−1, respectively. Furthermore, mCMC/Ge demonstrated efficient removal even in real water sample.The excellent removal performance of mCMC/Ge highlighted the synergy between polymeric matrix template and encapsulated Cit-Fe3O4 in improving Cipro adsorption and photodegradation. Furthermore, facile recyclability and sustained activity over five cycles identify mCMC/Ge photo-nanocomposite as a promising material for removing organic pollutants from contaminated waters.

Selectivity in Nontargeted Qualitative Analysis

BackgroundThe concept of selectivity is used extensively in analytical chemistry in both quantitative and qualitative analyses. While in the first case, the definition and use of selectivity have been somehow regulated, in the second case it remains a vague concept. (40) ResultsIn this paper we propose a new definition of selectivity in qualitative analysis as well as the way to estimate it numerically. (23) SignificanceIt is shown how selectivity can be used practically to compare analytical objects, measurement techniques, and data modeling methods. (19)

The FDA Sentinel Real World Evidence Data Enterprise (RWE-DE).

The US Food and Drug Administration's Sentinel Innovation Center aimed to establish a query-ready, quality-checked distributed data network containing electronic health records (EHRs) linked with insurance claims data for at least 10 million individuals to expand the utility of real-world data for regulatory decision-making. In this report, we describe the resulting network, the Real-World Evidence Data Enterprise (RWE-DE), including data from two commercial EHR-claims linked assets collectively termed the Commercial Network covering 21 million lives, and four academic partner institutions collectively termed the Development Network covering 4.5 million lives. We discuss provenance and completeness of the data converted in the Sentinel Common Data Model (SCDM), describe patient populations, and report on EHR-claims linkage characterization for all contributing data sources. Further, we introduce a standardized process to store free-text notes in the Development Network for efficient retrieval as needed. Finally, we outline typical use cases for the RWE-DE where it can broaden the reach of the types of questions that can be addressed by the Sentinel system.

A multi-modal EHR-based approach to screen patients for oncology clinical trials.

411 Background: Inefficiencies in patient (pt) screening may impair enrollment in oncology trials and could potentially be mitigated by a centralized trial matching infrastructure. The objective of this study was to implement a novel clinical trial screening service using electronic health record (EHR)-based structured variables, machine learning (ML) and human abstraction. Methods: To construct the model, we defined a set of inclusion and exclusion (IE) criteria unique to each trial and routinely documented in EHRs. Patients that met IE criteria via structured and ML-extracted clinical data were selected for review by trained specialists via a technology-enabled abstraction platform. Real-time notifications for eligible pts were issued via EHR alerts to research teams at participating study sites. Results: Between January and May 2024, over 80,000 pts were screened prior to a clinic visit for two trials at three sites. Using structured IE criteria alone, 98.7% and 99.6% of pts were excluded. An additional 31.9% of pts were excluded when employing ML modeling. A total of 15 (3.0%) and 5 (4.0%) pts were deemed eligible when using structured, ML, and abstraction-based criteria (Table). Seventeen percent of surfaced pts were consented within 10 days of implementation. Median time for abstractor confirmation of pt eligibility was 13.3 hours (interquartile range: 2.8, 27.4) with abstraction of highest priority pts taking a median of 3.0 and 9.9 hours, respectively. Conclusions: Our results support the implementation of a centralized, technology-enabled pt screening service for effective trial eligibility assessments across large patient populations. The value of using structured, ML- and abstraction-based data models, improved accuracy by more than 95% compared to structured data alone. The impact of multi-modal centralized screening is being evaluated across additional clinical trials in a nationwide community oncology research network. Pts, No. (% pts excluded). Trial 1 Trial 2 1. Pts with upcoming visit 37,865 44,973 2. Pts meeting structured criteria (including: age, stage, prior IV therapy) 497 (98.7) 182 (99.6) 3. Pts meeting ML model-based criteria (including: prior oral therapy) - 124 (31.9) 4. Pts meeting steps 1-3 and abstraction criteria (including: cancer diagnosis, biomarker status, prior malignancies) 15 (97.0) 5 (96.0)

Integration of Inventory Data from Cohort and Registry Studies into an Existing Research Network: National Pandemic Cohort Network (NAPKON)

In the early phase of the COVID-19 pandemic, many local collections of clinical data on patients infected with SARS-CoV-2 were initiated in Germany. As part of the National Pandemic Cohort Network (NAPKON) of the University Medicine Network, the "Integration Core" was established to design the legal, technical and organisational requirements for the integration of inventory data into ongoing prospective data collections and to test the feasibility of the newly developed solutions using use cases (UCs). Detailed study documents of the data collections were obtained. After structured document analysis, a review board evaluated the integrability of the data in NAPKON according to defined criteria. Of 30 university hospitals contacted, 20 responded to the request. Patient information and consent showed a heterogeneous picture with regard to the pseudonymised transfer of data to third parties and re-contact. The majority of the data collections (n=13) met the criteria for integration into NAPKON; four studies would require adjustments to the regulatory documents. Three cohorts were not suitable for inclusion in NAPKON. The legal framework for retrospective data integration and consent-free data use via research clauses (§27 BDSG) was elaborated by a legal opinion by TMF - Technology, Methods and Infrastructure for Networked Medical Research, Berlin. Two UCs selected by the NAPKON steering committee (CORKUM, LMU Munich; Pa-COVID-19, Charité- Universitätsmedizin Berlin) were used to demonstrate the feasibility of data integration in NAPKON by the end of 2021. Quality assurance and performance-based reimbursement of the cases were carried out according to the specifications. Based on the results, recommendations can be formulated for various contexts in order to create technical-operational prerequisites such as interoperability, interfaces and data models for data integration and to fulfil regulatory requirements on ethics, data protection, medical confidentiality and data access when integrating existing cohort data. The possible integration of data into research networks and their secondary use should be taken into account as early as the planning phase of a study - particularly with regard to informed consent - in order to maximise the benefits of the data collected.

The use of predictive modelling to determine the likelihood of donor return during the COVID-19 pandemic.

Artificial intelligence (AI) uses sophisticated algorithms to "learn" from large volumes of data. This could be used to optimise recruitment of blood donors through predictive modelling of future blood supply, based on previous donation and transfusion demand. We sought to assess utilisation of predictive modelling and AI blood establishments (BE) and conducted predictive modelling to illustrate its use. A BE survey of data modelling and AI was disseminated to the International Society of Blood transfusion members. Additional anonymzed data were obtained from Italy, Singapore and the United States (US) to build predictive models for each region, using January 2018 through August 2019 data to determine likelihood of donation within a prescribed number of months. Donations were from March 2020 to June 2021. Ninety ISBT members responded to the survey. Predictive modelling was used by 33 (36.7%) respondents and 12 (13.3%) reported AI use. Forty-four (48.9%) indicated their institutions do not utilise predictive modelling nor AI to predict transfusion demand or optimise donor recruitment. In the predictive modelling case study involving three sites, the most important variable for predicting donor return was number of previous donations for Italy and the US, and donation frequency for Singapore. Donation rates declined in each region during COVID-19. Throughout the observation period the predictive model was able to consistently identify those individuals who were most likely to return to donate blood. The majority of BE do not use predictive modelling and AI. The effectiveness of predictive model in determining likelihood of donor return was validated; implementation of this method could prove useful for BE operations.

PENGARUH GOOD CORPORATE GOVERNANCE DAN GREEN INNOVATION TERHADAP KINERJA BERKELANJUTAN PERUSAHAAN

This study aims to analyze the effect of good corporate governance and green innovation on sustainability performance. This study is classified as an associative quantitative study. The type of data used is secondary data obtained from www.idx.co.id and the company's website. The population in this study is the Sri-Kehati Company Listed on the IDX for the 2019-2023 Period. While the sample of this study was determined by the purposive sampling method so that 20 sample companies were obtained. The analysis method used is Panel Data Model Regression analysis. The results of this study indicate that good corporate governance and green innovation have an effect on sustainability performance, good corporate governance has an effect on sustainability performance, green innovation has no effect on sustainability performance.

CSEM Optimization Using the Correspondence Principle

Traditionally, 3D modeling of marine controlled-source electromagnetic (CSEM) data (in the frequency domain) involves high-memory demand, requiring solving a large linear system for each frequency. To address this problem, we propose to solve Maxwell’s equations in a fictitious dielectric medium with time-domain finite-difference methods, with the support of the correspondence principle. As an advantage of this approach, we highlight the possibility of its implementation for execution with GPU accelerators, in addition to multi-frequency data modeling with a single simulation. Furthermore, we explore using the correspondence principle to the inversion of CSEM data by calculating the gradient of the least-squares objective function employing the adjoint-state method to establish the relationship between adjoint fields in a conductive medium and their counterparts in the fictitious dielectric medium, similar to the approach used in forward modeling. We validate this method through 2D inversions of three synthetic CSEM datasets, computed for a simple model consisting of two resistors in a conductive medium, a model adapted from a CSEM modeling and inversion package, and the last one based on a reference model of turbidite reservoirs on the Brazilian continental margin. We also evaluate the differences between the results of inversions using the steepest descent method and our proposed momentum method, comparing them with the limited-memory BFGS (Broyden–Fletcher–Goldfarb–Shanno) algorithm (L-BFGS-B). In all experiments, we use smoothing by model reparameterization as a strategy for regularizing and stabilizing the iterations throughout the inversions. The results indicate that, although it requires more iterations, our modified momentum method produces the best models, which are consistent with results from the L-BFGS-B algorithm and require less storage per iteration.