High-quality Database Research Articles

Machine-learning based prediction of obstructive coronary artery disease using integrated submodules of clinical information, chest x-ray, and electrocardiography

Abstract Background The early diagnosis of obstructive coronary artery disease (CAD) is critical. However, delays in diagnosis may occur due to subsequent high-cost tests, contributing to substantial medical expenses. A more insightful interpretation of fundamental diagnostic tools, such as chest x-ray (CXR) and electrocardiography (ECG), could mitigate medical costs and facilitate a timely diagnosis. Purpose This study explores the application of machine learning to interpret each modality and create a consolidated prediction model for significant CAD, aiming to construct a diagnostic model based on primary tests, including CXR, ECG, and clinical information. Methods Clinical information-ECG-CXR paired data were generated from 19,140 patients, with CAD confirmed through coronary angiography (CAG). Machine learning (ML) was employed to develop submodules for each modality, and various integration methods were explored. The area under the curve (AUC) served as the outcome metric for predicting obstructive CAD. Results The development set comprised data from 17,976 patients, with CAG-confirmed obstructive CAD observed in approximately 60% of patients. The obstructive CAD group exhibited characteristics such as advanced age, male predominance, a higher prevalence of chest pain, and more risk factors. The ML model, incorporating clinical information, CXR, and ECG with submodules, demonstrated the optimal prediction of obstructive CAD (AUC 0.722). This model significantly outperformed the model without submodules (0.722 vs. 0.638, p<0.0001) in obstructive CAD prediction. Conclusions Through the integration of submodules encompassing clinical information and basic tests, leveraging a high-quality database, the integrated ML algorithm offers improved prediction capabilities for CAG-confirmed obstructive CAD. This underscores the potential role of machine learning in clinical decision-making based on diverse modalities with distinct characteristics.

Unfolding wall turbulence

We present the project FLARE (Unfolding wall turbulence), funded by the Spanish Ministry of Science and FEDER. The main objective of this project is to contribute to understanding the dynamics of wall-bounded turbulence, exploring two techniques: Lie-Symmetry and Machine Learning. Further, we have compiled several high-quality databases of different flows using Direct Numerical Simulations of the Navier-Stokes equations with the code LISO to use these techniques. This data is available to the turbulence community.

DC_OCEAN: an open-source algorithm for identification of duplicates in ocean databases

A high-quality hydrographic observational database is essential for ocean and climate studies and operational applications. Because there are numerous global and regional ocean databases, duplicate data continues to be an issue in data management, data processing and database merging, posing a challenge on effectively and accurately using oceanographic data to derive robust statistics and reliable data products. This study aims to provide algorithms to identify the duplicates and assign labels to them. We propose first a set of criteria to define the duplicate data; and second, an open-source and semi-automatic system to detect duplicate data and erroneous metadata. This system includes several algorithms for automatic checks using statistical methods (such as Principal Component Analysis and entropy weighting) and an additional expert (manual) check. The robustness of the system is then evaluated with a subset of the World Ocean Database (WOD18) with over 600,000 in-situ temperature and salinity profiles. This system is an open-source Python package (named DC_OCEAN) allowing users to effectively use the software. Users can customize their settings. The application result from the WOD18 subset also forms a benchmark dataset, which is available to support future studies on duplicate checks, metadata error identification, and machine learning applications. This duplicate checking system will be incorporated into the International Quality-controlled Ocean Database (IQuOD) data quality control system to guarantee the uniqueness of ocean observation data in this product.

Fruchtbringende Gesellschaft (1617–1680) Member Publication Patterns in the VD17

This paper presents the first results of a larger project that draws on large-scale data analysis to investigate the publication patterns and networks of the 890 members of the Fruchtbringende Gesellschaft (1617–1680), or Fruitbearing Society, the first and largest cultural society in early modern Central Europe. First, we elucidate the major steps, including data wrangling, evaluation, clean-up, and algorithmic enrichment, necessary to transform the already high-quality VD17 bibliographic database into research data. Then, we relate the first results of our investigation of the publication patterns of Society members over time. This analysis brings more nuance to the existing narrative of a society that shifted from focusing on the literary and linguistic aspects of its agenda in the first period (1617–1650) to a more courtly one in the later periods (1651–1662/67) of its existence

The role of human errors and violations in pedestrian-related crashes: Harnessing a unique database and accounting for heterogeneity

IntroductionHuman factors are often major contributors to pedestrian crashes. However, police-reported pedestrian-involved crash data often have gaps in crash details. Overcoming this limitation, the Pedestrian and Bicycle Crash Data Tool (PBCAT) provides a more comprehensive high-quality database capturing the sequence of events. MethodsIn addition to human and roadway environmental factors, there could be unobserved factors (e.g., pedestrian conspicuity, impact speed, or riskiness by a driver) that could be either unavailable or not used in the analysis; however, these unobserved factors could have a significant influence on pedestrian injuries. This study applies finite mixture models to address unobserved heterogeneity in pedestrian injuries which is usually overlooked. As a result, the associations of one or more of the observed factors with pedestrian injuries across different latent (unobserved) classes can be different. ResultsHarnessing this unique database for North Carolina reveals that in most (95%) of the crashes (N=24,886) occurring between 2009 and 2019, pedestrians were either killed or injured. Risky behaviors by drivers and pedestrians contributed to 7.91% and 50.59% of these crashes, respectively. Recognition errors (e.g., dash or dart-out) and violations (e.g., failure to yield) by pedestrians contributed to 22.08% and 28.58% of crashes, respectively. Recognition errors and violations by drivers contributed to only 2.12% and 3.11% of crashes respectively each of which is significantly lower than those by pedestrians. Results of the ordered Probit model indicate that the chance of pedestrian fatality is significantly higher if a pedestrian makes recognition errors and violations, a driver makes performance errors, and either the pedestrian or driver is impaired. Conclusions and practical implicationsThe finite mixture model shows that pedestrians belong to two latent groups across which there is significant heterogeneity in pedestrian injuries and variations in the associations of observed factors with pedestrian injuries. The practical implications are discussed in the paper.

Multi-modal deep learning enables efficient and accurate annotation of enzymatic active sites

Annotating active sites in enzymes is crucial for advancing multiple fields including drug discovery, disease research, enzyme engineering, and synthetic biology. Despite the development of numerous automated annotation algorithms, a significant trade-off between speed and accuracy limits their large-scale practical applications. We introduce EasIFA, an enzyme active site annotation algorithm that fuses latent enzyme representations from the Protein Language Model and 3D structural encoder, and then aligns protein-level information with the knowledge of enzymatic reactions using a multi-modal cross-attention framework. EasIFA outperforms BLASTp with a 10-fold speed increase and improved recall, precision, f1 score, and MCC by 7.57%, 13.08%, 9.68%, and 0.1012, respectively. It also surpasses empirical-rule-based algorithm and other state-of-the-art deep learning annotation method based on PSSM features, achieving a speed increase ranging from 650 to 1400 times while enhancing annotation quality. This makes EasIFA a suitable replacement for conventional tools in both industrial and academic settings. EasIFA can also effectively transfer knowledge gained from coarsely annotated enzyme databases to smaller, high-precision datasets, highlighting its ability to model sparse and high-quality databases. Additionally, EasIFA shows potential as a catalytic site monitoring tool for designing enzymes with desired functions beyond their natural distribution.

Accuracy Analysis of Topographical and Morphological Code in the RSUP Dr. Sardjito Cancer Registry, Indonesia

In cancer cases, a cancer registry functions as a national cancer control program to analyze cancer severity and risk factors in specific areas, aiming to reduce morbidity and mortality. Accurate coding is essential for producing a high-quality cancer database. This study aims to analyze the accuracy of the cancer registry at RSUP Dr. Sardjito in assigning topography and morphology codes. A mixed-methods approach was employed to evaluate code accuracy and identify factors influencing accuracy and inaccuracy based on the 5M framework. Data were sourced from the Hospital-Based Cancer Registry at RSUP Dr. Sardjito, with a total population of 4500 patients in 2021. A sample of 1000 cancer patients was selected using a simple random sampling technique. Each sample was analyzed as true or false based on verifier notes, revealing that 904 topography codes were accurate (90.40%) and 96 were inaccurate (9.60%). For morphology codes, 913 were accurate (91.30%) and 87 were inaccurate (8.70%). Researchers conducted in-depth interviews with registrars and verifier doctors and observed all cancer registry activities, focusing on topography and morphology codification. Observations and interviews indicated no obstacles in codification activities, but inaccuracies occurred, possibly due to case complexity. In conclusion, the study demonstrates high accuracy in the coding of cancer cases at RSUP Dr. Sardjito. The complexity of cases may contribute to inaccuracies. The study suggests continuous evaluation to enhance coding accuracy, which is crucial for effective cancer control programs.

Assessment of performance of the profilers provided in the OECD QSAR toolbox for category formation of chemicals

Growing restrictions and bans on animal testing for chemical safety assessment under different regulations have led to an increasing use of alternative methods. Read-across is one of the major approaches used for this purpose, which relies on the identification of toxicological hazards of a data-poor or untested (target) chemical from data on other already-tested (source) similar chemicals. This requires the target substance to be first assigned to a group or category of ‘similar’ chemicals. The ‘similarity’ may be in terms of structural features alone, or in combination with certain rules that are based on mechanistic and/or toxicological aspects. In this regard, the OECD QSAR toolbox—a major free-access in silico platform—is widely used to derive toxicity predictions for a range of (eco) toxicological endpoints. The Toolbox allows the user to identify a set of similar chemicals (analogues) by computational ‘profilers’ that incorporate different structural alerts, or a combination of structural alerts and physicochemical and/or toxicokinetic rules relevant to a specific toxicological endpoint. The overall aim of this study was to assess the performance of the in silico profilers provided in the OECD QSAR Toolbox for reliability for identifying chemical analogues for category formation in a number of high-quality databases on mutagenicity, carcinogenicity, and skin sensitisation. The study also aimed to identify the reasons for any limitations in the performance of the profilers, and propose ways to improve their overall accuracy. The results showed that whilst some structural alerts are fit-for-purpose as such within the acceptable limits, others need refinement or a consideration for their possible exclusion from the profiler. Such refinements are imperative for a reliable use of the profilers in read-across and grouping/categorisation for classification, labelling and risk assessment of chemicals.

Psychological and behavior investigation of Chinese residents: Concepts, practices, and prospects

To provide researchers with a comprehensive understanding of the Psychological and Behavior Investigation of Chinese Residents (PBICR), this paper introduces its background and features from four aspects. In terms of design concept, the PBICR focuses on various aspects of the mental health and health behaviors of the Chinese population, takes full account of timeliness and efficiency, adheres to the original intention of open data access and data sharing, and builds a high-quality database with large-sample, multi-center, repetitive, and a nationwide cross-sectional design to promote data mining and exchange and cooperation; in terms of survey implementation, the PBICR adopts a combination of population size-proportional, multi-stage sampling and quota sampling, and obtains data through face-to-face field surveys with strict quality control to ensure the representativeness and reliability of the samples. Regarding to the output, the PBICR's research content is rich and updated in line with international hotspots, which can satisfy the demand for research needs on diversified variables and data. Previous research results have had a significant impact in many fields such as public health, management, communication, and psychology. Looking ahead, the PBICR will gradually complete the construction of databases from the general database to sub-databases, pay attention to special populations and globally widespread areas, add perspectives on tracking surveys and biomedical data research, and have greater research potential to drive research on the mental and behavioral health of Chinese population through the overall layout of the multi-dimensional.

Prediction of Drug-Induced Liver Injury: From Molecular Physicochemical Properties and Scaffold Architectures to Machine Learning Approaches.

The process of developing new drugs is widely acknowledged as being time-intensive and requiring substantial financial investment. Despite ongoing efforts to reduce time and expenses in drug development, ensuring medication safety remains an urgent problem. One of the major problems involved in drug development is hepatotoxicity, specifically known as drug-induced liver injury (DILI). The popularity of new drugs often poses a significant barrier during development and frequently leads to their recall after launch. In silico methods have many advantages compared with traditional invivo and invitro assays. To establish a more precise and reliable prediction model, it is necessary to utilize an extensive and high-quality database consisting of information on drug molecule properties and structural patterns. In addition, we should also carefully select appropriate molecular descriptors that can be used to accurately depict compound characteristics. The aim of this study was to conduct a comprehensive investigation into the prediction of DILI. First, we conducted a comparative analysis of the physicochemical properties of extensively well-prepared DILI-positive and DILI-negative compounds. Then, we used classic substructure dissection methods to identify structural pattern differences between these two different types of chemical molecules. These findings indicate that it is not feasible to establish property or substructure-based rules for distinguishing between DILI-positive and DILI-negative compounds. Finally, we developed quantitative classification models for predicting DILI using the naïve Bayes classifier (NBC) and recursive partitioning (RP) machine learning techniques. The optimal DILI prediction model was obtained using NBC, which combines 21 physicochemical properties, the VolSurf descriptors and the LCFP_10 fingerprint set. This model achieved a global accuracy (GA) of 0.855 and an area under the curve (AUC) of 0.704 for the training set, while the corresponding values were 0.619 and 0.674 for the test set, respectively. Moreover, indicative substructural fragments favorable or unfavorable for DILI were identified from the best naïve Bayesian classification model. These findings may help prioritize lead compounds in the early stage of drug development pipelines.

Database for the Fault Displacement Hazard Initiative Project

New predictive models for fault displacement and surface rupture hazard analysis developed through the Fault Displacement Hazard Initiative (FDHI) research program require a high-quality empirical database to apply advanced statistical methods and improve hazard estimates. This article discusses the development and contents of the FDHI Project database. We systematically collected, reviewed, and organized fault displacement measurements, surface rupture maps, and supporting information from the scientific literature. A framework was developed and implemented to classify principal and distributed faulting. Best-estimate net displacement amplitudes were calculated from slip component measurements and quality codes were assigned to all net displacement values. The database contains 75 historical, surface-rupturing crustal earthquakes ranging from M 4.9 to 8.0. Thirty-five earthquakes have a strike-slip faulting mechanism, while 25 and 15 events are reverse/reverse-oblique and normal/normal-oblique, respectively. Although most of the earthquakes are from Western North America, Japan, and other active tectonic regions, there are nine reverse faulting events from the stable continental region of Australia. The database contains over 40,000 individual fault displacement measurements for various slip components from roughly 28,000 observation sites. Geographic coordinates are included for all data, and event-specific coordinate systems are provided for each earthquake that transform data into an along-strike dimension. Our new database provides a standardized collection of surface rupture and fault displacement data and metadata that is the result of a comprehensive effort to create a reliable and stable product for the FDHI model development teams and the geoscience community.

Real-time confinement regime detection in fusion plasmas with convolutional neural networks and high-bandwidth edge fluctuation measurements

A real-time detection of the plasma confinement regime can enable new advanced plasma control capabilities for both the access to and sustainment of enhanced confinement regimes in fusion devices. For example, a real-time indication of the confinement regime can facilitate transition to the high-performing wide-pedestal (WP) quiescent H-mode, or avoid unwanted transitions to lower confinement regimes that may induce plasma termination. To demonstrate real-time confinement regime detection, we use the 2D beam emission spectroscopy (BES) diagnostic system to capture localized density fluctuations of long wavelength turbulent modes in the edge region at a 1 MHz sampling rate. BES data from 330 discharges in either L-mode, H-mode, quiescent H (QH)-mode, or WP QH-mode were collected from the DIII-D tokamak and curated to develop a high-quality database to train a deep-learning classification model for real-time confinement detection. We utilize the 6×8 spatial configuration with a time window of 1024 µs and recast the input to obtain spectral-like features via fast Fourier transform preprocessing. We employ a shallow 3D convolutional neural network for the multivariate time-series classification task and utilize a softmax in the final dense layer to retrieve a probability distribution over the different confinement regimes. Our model classifies the global confinement state on 44 unseen test discharges with an average F 1 score of 0.94, using only ∼1 ms snippets of BES data at a time. This activity demonstrates the feasibility for real-time data analysis of fluctuation diagnostics in future devices such as ITER, where the need for reliable and advanced plasma control is urgent.

SHAPEwarp-web: sequence-agnostic search for structurally homologous RNA regions across databases of chemical probing data.

RNA molecules perform a variety of functions in cells, many of which rely on their secondary and tertiary structures. Chemical probing methods coupled with high-throughput sequencing have significantly accelerated the mapping of RNA structures, and increasingly large datasets of transcriptome-wide RNA chemical probing data are becoming available. Analogously to what has been done for decades in the protein world, this RNA structural information can be leveraged to aid the discovery of structural similarity to a known RNA (or RNA family), which, in turn, can inform about the function of transcripts. We have previously developed SHAPEwarp, a sequence-agnostic method for the search of structurally homologous RNA segments in a database of reactivity profiles derived from chemical probing experiments. In its original implementation, however, SHAPEwarp required substantial computational resources, even for moderately sized databases, as well as significant Linux command line know-how. To address these limitations, we introduce here SHAPEwarp-web, a user-friendly web interface to rapidly query large databases of RNA chemical probing data for structurally similar RNAs. Aside from featuring a completely rewritten core, which speeds up by orders of magnitude the search inside large databases, the web server hosts several high-quality chemical probing databases across multiple species. SHAPEwarp-web is available fromhttps://shapewarp.incarnatolab.com.

POSTER AS A TOOL FOR DIGITALIZATION OF THE MANAGEMENT SYSTEM OF HOTEL AND RESTAURANT BUSINESS ENTERPRISES

In the basis of the development of modern business, digital transformations have acquired primary importance. The stability of the business, and in general its ability to function, is determined by the level of digitalization. The integration of basic business development tools is considered through the prism of automation, robotics and digitization. In the field of hotel and restaurant business, digital technologies have become the most effective tool for implementing business tasks and solving problems of the control system, constant monitoring and management. The main trajectory of the automation of hospitality enterprises through the prism of the introduction of cloud technologies into business has become the use of software products. Cloud automation systems, in particular the Poster program, serve as a modern tool for accounting, control and business management. The Poster program (system) was the first cloud-based automation system developed in Ukraine to serve hotel and restaurant businesses. This software product has become a successful modern initiative, which is used with pleasure by twenty thousand hospitality establishments. When the enterprise implements this cloud technology, the approach to the organization and conduct of business has completely changed, the management system has acquired a completely new dimension. Important functions of the module have become such as: cash register for waiters with integrated hall cards; the Kitchen Kit application from Poster with a screen in the kitchen/bar, or with a system of runners; automation of the restaurant delivery service to work with orders from aggregators or personal delivery service; Poster mobile cash register; Poster QR for automatic menu creation in Poster-POS, quick payment, the possibility of tipping the waiter, leaving guest reviews, etc.; the Poster Boss app with the ability to work on a smartphone and more. The convenience of working with the Poster software is manifested in its simplicity and easy accessibility for staff training, since even intuitively, in an hour of work with this cloud technology, you can already work freely. The ability to work with cash register modules for sales and a control panel for owners makes the Poster software product both unique due to its multifunctionality and system, and easy to use. The statistics module automatically summarizes revenue data when the check is closed at the cash register. In the process of selling a finished product or service, ingredients (inventories) are written off from the warehouse according to technological maps developed by the institution. The multifunctionality of the Poster program makes running a hotel and restaurant business more efficient, as convenient as possible in management and acts as a high-quality database for making effective management decisions. The complexity and systematicity of this cloud technology brings the innovative orientation of hospitality enterprises to a new level.

Artificial intelligence in maxillofacial and facial plastic and reconstructive surgery.

To provide a current review of artificial intelligence and its subtypes in maxillofacial and facial plastic surgery including a discussion of implications and ethical concerns. Artificial intelligence has gained popularity in recent years due to technological advancements. The current literature has begun to explore the use of artificial intelligence in various medical fields, but there is limited contribution to maxillofacial and facial plastic surgery due to the wide variance in anatomical facial features as well as subjective influences. In this review article, we found artificial intelligence's roles, so far, are to automatically update patient records, produce 3D models for preoperative planning, perform cephalometric analyses, and provide diagnostic evaluation of oropharyngeal malignancies. Artificial intelligence has solidified a role in maxillofacial and facial plastic surgery within the past few years. As high-quality databases expand with more patients, the role for artificial intelligence to assist in more complicated and unique cases becomes apparent. Despite its potential, ethical questions have been raised that should be noted as artificial intelligence continues to thrive. These questions include concerns such as compromise of the physician-patient relationship and healthcare justice.

Combined use of Donepezil and Memantine increases the probability of five-year survival of Alzheimer’s disease patients

BackgroundAlzheimer’s disease (AD) is the most common neurodegenerative disease. Studying the effects of drug treatments on multiple health outcomes related to AD could be beneficial in demonstrating which drugs reduce the disease burden and increase survival.MethodsWe conducted a comprehensive causal inference study implementing doubly robust estimators and using one of the largest high-quality medical databases, the Oracle Electronic Health Records (EHR) Real-World Data. Our work was focused on the estimation of the effects of the two common Alzheimer’s disease drugs, Donepezil and Memantine, and their combined use on the five-year survival since initial diagnosis of AD patients. Also, we formally tested for the presence of interaction between these drugs.ResultsHere, we show that the combined use of Donepezil and Memantine significantly elevates the probability of five-year survival. In particular, their combined use increases the probability of five-year survival by 0.050 (0.021, 0.078) (6.4%), 0.049 (0.012, 0.085), (6.3%), 0.065 (0.035, 0.095) (8.3%) compared to no drug treatment, the Memantine monotherapy, and the Donepezil monotherapy respectively. We also identify a significant beneficial additive drug-drug interaction effect between Donepezil and Memantine of 0.064 (0.030, 0.098).ConclusionsBased on our findings, adopting combined treatment of Memantine and Donepezil could extend the lives of approximately 303,000 people with AD living in the USA to be beyond five-years from diagnosis. If these patients instead have no drug treatment, Memantine monotherapy or Donepezil monotherapy they would be expected to die within five years.

LabVIEW-Enabled Synthetic Signal for Empowering Fetal-Maternal Healthcare

Biomedical signal processing has advanced to the point that tools and methods are now available to doctors to diagnose and track medical conditions connected to pregnancy. However, it is extremely difficult for researchers to look into novel procedures and approaches to uncover underlying pathological abnormalities associated with high-risk pregnancies due to the scarcity of high-quality medical databases of pregnant women. In this study, a LabVIEW software environment is used to precisely design a bio-physiological signal generator (BPSG) for use in feto-maternal health assessment applications. McSharry’s dynamical ECG model served as inspiration for the methods utilized to create the proposed time-domain mathematical model. The BPSG is capable of generating various realistic synthetic signals like respiration signal, pulse plethysmography (PPG) signal, phonocardiography (PCG) signal, maternal ECG (MECG) signal, fetal ECG (FECG) signal, abdominal ECG (AECG) signa,l and umbilical blood flow (UBF) velocimetry signals with corresponding Doppler indices. It is possible to create synthetic signals for both healthy and unhealthy conditions. Synthetic signal facilitates the testing and calibration of new diagnostic procedures, denoising algorithms, feature extraction processes, and instrumentation, all of which contribute to the prompt prediction of an overall health state of expectant mother.

QuanDB: a quantum chemical property database towards enhancing 3D molecular representation learning

Previous studies have shown that the three-dimensional (3D) geometric and electronic structure of molecules play a crucial role in determining their key properties and intermolecular interactions. Therefore, it is necessary to establish a quantum chemical (QC) property database containing the most stable 3D geometric conformations and electronic structures of molecules. In this study, a high-quality QC property database, called QuanDB, was developed, which included structurally diverse molecular entities and featured a user-friendly interface. Currently, QuanDB contains 154,610 compounds sourced from public databases and scientific literature, with 10,125 scaffolds. The elemental composition comprises nine elements: H, C, O, N, P, S, F, Cl, and Br. For each molecule, QuanDB provides 53 global and 5 local QC properties and the most stable 3D conformation. These properties are divided into three categories: geometric structure, electronic structure, and thermodynamics. Geometric structure optimization and single point energy calculation at the theoretical level of B3LYP-D3(BJ)/6-311G(d)/SMD/water and B3LYP-D3(BJ)/def2-TZVP/SMD/water, respectively, were applied to ensure highly accurate calculations of QC properties, with the computational cost exceeding 107 core-hours. QuanDB provides high-value geometric and electronic structure information for use in molecular representation models, which are critical for machine-learning-based molecular design, thereby contributing to a comprehensive description of the chemical compound space. As a new high-quality dataset for QC properties, QuanDB is expected to become a benchmark tool for the training and optimization of machine learning models, thus further advancing the development of novel drugs and materials. QuanDB is freely available, without registration, at https://quandb.cmdrg.com/.

A spatiotemporal feature fusion-based deep learning framework for synchronous prediction of excavation stability

Under rapid urbanization, analyzing and perceiving synchronous excavation stability is essential to generate valuable and foresightful information in identifying and mitigating potential risks in deep excavations. Although the existing forecasting methods can achieve high accuracy in the safety state perception of deep excavations, it is prone to lack proper consideration of the monitoring item’s fusion and cause error and uncertainty via the single-parameter risk evaluation. To address it, this paper proposes a spatiotemporal feature fusion-based deep learning (FFDL) framework to implement the feature fusion-based data analysis and time-series forecasting on the factor of safety (FS) on behalf of excavation stability, and global sensitivity analysis. The fusion indicators and proposed FFDL framework are validated through a comprehensive case study encompassing 100 typical metro station constructions. The feature fusion-based data analysis results confirm the statistical significance of fusion indicators, contributing to establishing a high-quality spatiotemporal sequential database. As for the real-time prediction, the Bayesian optimization CNN-LSTM model provides a reliable prediction of excavation stability and the relative errors in the training and testing datasets do not exceed 1% and 3% in construction duration. It also enhances the predictive accuracy and outperforms the existing models LSTM, GRU, BiLSTM, and CNN-LSTM, reaching an average improvement of 38%, 32%, and 32% in RMSE, MAE, and MAPE, respectively. Furthermore, the proposed model has an excellent generalization performance in accurately predicting the excavation stability of the other ten excavation cases with an average relative error of 5.37%. In short, this work contributes to not only providing a new perspective in risk management by integrating feature fusion-based indicators and excavation stability but also verifying the applicability of the proposed framework.

Indoor positioning fingerprint database construction based on CSA-DBSCAN and RCVAE-GAN

With the increasing size of buildings, in order to achieve high-precision indoor positioning services, it is a challenging task to build an offline fingerprint database with high quality, high density and less manpower and material consumption. Aiming to solve the problem of low-quality WiFi indoor positioning fingerprint inventory constructed by traditional methods, which affects positioning accuracy and incurs high costs, this paper proposes a method for indoor positioning fingerprint database construction based on Crow Search Algorithm Optimizes Density Clustering (CSA-DBSCAN) and Regressor Conditional VAE Generative Adversarial Network (RCVAE-GAN). Collecting only a tiny amount of sparse reference point position coordinates and RSS data makes it possible to construct a high-quality WiFi indoor positioning fingerprint database. Firstly, the method utilizes the density clustering method based on Crow Search Algorithm Optimization (CSA-DBSCAN) to process RSS data collected from the reference point. This helps minimize the impact of abnormal RSS data on creating the fingerprint database. Secondly, the RCVAE-GAN depth generation model was developed. The model consists of an encoder E, a generator G, a discriminator D, and a regressor R. After constructing the model, the data with abnormal RSS will be removed and input into the model for pre-training and joint training, resulting in a high-quality deep-generation model. Finally, a high-quality and high-density fingerprint database is constructed by combining the collected reference points with fingerprint data generated by the depth generation model. Experimental results show that the proposed method reduces the root mean square error (RMSE) deviation of the generated fingerprint data by 38% and 12% respectively, compared to the RBF interpolation method and the CVAE-GAN method in the same experimental scenario. The constructed fingerprint database is used for positioning, improving positioning accuracy by 70% and 65% respectively. The method described in this paper can construct a high-quality fingerprint database, effectively improving the efficiency of fingerprint database construction and reducing the costs associated with labor and time.