Multiple Data Sources Research Articles

DORA: an interactive map for the visualization and analysis of ancient human DNA and associated data.

The ability to sequence ancient genomes has revolutionized the way we study evolutionary history by providing access to the most important aspect of evolution-time. Until recently, studying human demography, ecology, biology, and history using population genomic inference relied on contemporary genomic datasets. Over the past decade, the availability of human ancient DNA (aDNA) has increased rapidly, almost doubling every year, opening the way for spatiotemporal studies of ancient human populations. However, the multidimensionality of aDNA, with genotypes having temporal, spatial and genomic coordinates, and integrating multiple sources of data, poses a challenge for developing meta-analyses pipelines. To address this challenge, we developed a publicly-available interactive tool, DORA, which integrates multiple data types, genomic and non-genomic, in a unified interface. This web-based tool enables browsing sample metadata alongside additional layers of information, such as population structure, climatic data, and unpublished samples. Users can perform analyses on genotypes of these samples, or export sample subsets for external analyses. DORA integrates analyses and visualizations in a single intuitive interface, resolving the technical issues of combining datasets from different sources and formats, and allowing researchers to focus on the scientific questions that can be addressed through analysis of aDNA datasets.

Breast cancer and the steadily increasing maternal age: are they colliding?

BackgroundPregnancy-related cancers are mostly breast cancers, and their incidence is likely to increase as a result of the modern trend of delaying childbearing. In particular, advanced maternal age increases breast cancer risk, and younger breast cancer patients are more likely to die and metastasize. This study compared a population with a high incidence of delayed childbearing with another population with a lower mean age at childbirth in order to determine whether breast cancer diagnosis and childbearing age overlap.MethodsWe retrospectively analyzed multiple data sources. The Surveillance, Epidemiology, and End Results (SEER) program, the United States National Center for Health Statistics as part of the National Vital Statistics System, the United Nations Population Division, the GLOBOCAN Cancer Observatory, the CLIO-INFRA project database, the Human Fertility Database, and anonymized local data were used.ResultsAs women’s age at delivery increased, the convergence between their age distribution at breast cancer diagnosis and childbearing increased. In addition, the overlap between the two age distributions increased by more than 200% as the average age at delivery increased from 27 to 35 years.ConclusionsAs women’s average childbearing age has progressively risen, pregnancy and breast cancer age distributions have significantly overlapped. This finding emphasizes the need for increased awareness and educational efforts to inform women about the potential consequences of delayed childbearing. By providing comprehensive information and support, women can make more informed decisions about their reproductive health and cancer prevention strategies.

Fostering Privacy in Collaborative Data Sharing via Auto-encoder Latent Space Embedding

Securing privacy in machine learning via collaborative data sharing is essential for organizations seeking to harness collective data while upholding confidentiality. This becomes especially vital when protecting sensitive information across the entire machine learning pipeline, from model training to inference. This paper presents an innovative framework utilizing Representation Learning via autoencoders to generate privacy-preserving embedded data. As a result, organizations can distribute these representations, enhancing the performance of machine learning models in situations where multiple data sources converge for a unified predictive task downstream.

*READ**THIS*!! Spam as a threat for open science

Drawing on multiple sources of qualitative data, I describe a case of open science infrastructure (OSI) abuse. The case illustrates how developers navigated scholarly value tensions and issues of epistemic and platform legitimacy while battling spam on their open science webapp. Notably, their struggle used precious financial resources and drew attention away from other development tasks like feature expansion. This research makes evident that not only is OSI abuse like spam a financial burden, but it puts scholarly information security—specifically, the legitimacy of open science content—at risk. However, protecting against such abuse is not a trivial matter; it raises questions of who is responsible for defining and enforcing scholarly values. The urgency of this issue is magnified by OSI’s relationship to public trust in science.

Variations in the coherence and engagement in students’ experience of blended learning

We report a study which examines coherence of engagement of 344 first year engineering students’ blended learning experience. Using self-report and observational data sources, we demonstrate that student perceptions of the blended learning environment, academic learning outcomes, and engagement with the online learning activities are logically related at the variable level as shown by correlation analyses; and at the level of student groupings of similar learning experience and behaviors, as revealed by cluster, ANOVA, and 2 x 2 contingency analyses. Using self-report data, we found that when students perceived the learning activities in the f2f and online environments were integrated, they were more likely to be engaged with the online learning and to perform relatively higher on the assessment tasks than students who perceived disintegration between f2f and online learning. Using the observational data, students who were more engaged with the online learning tended to perceive that the online learning was well integrated with the f2f learning, that the online contributions were valuable for the whole learning experience, and achieved relatively higher than less engaged students. A 2 x 2 contingency table revealed a logical relationship between the groupings of students based on the selfreport and observational data: moderate and positive association was found between students with coherent perceptions and more engagement; and between students with fragmented perceptions and less engagement. The use of multiple data sources and methods enabled triangulation, strengthened analysis power, and offered a more comprehensive picture of students’ blended learning experience.

“Image-Spectral” fusion monitoring of small cotton samples nitrogen content based on improved deep forest

Accurate monitoring of nitrogen nutrition is crucial for improving cotton yield and quality, as well as the ecological environment. The mainstream method for monitoring nutrition is to establish traditional machine learning (ML) models using a single data source. However, this approach has limitations such as limited access to feature information, model-fitting problems, and limited generalization. Deep learning (DL), on the other hand, has shown promise in complex nonlinear modeling tasks due to its flexible structure. However, it has its limitations, such as the fact that agronomic sample collection and testing are usually labor- and material-intensive, resulting in sample sizes that are too small to meet its training conditions. Therefore, there is an urgent need for DL models that can effectively integrate features from multiple data sources and accurately monitor crop nitrogen content, especially in scenarios with small samples. In this study, we conducted indoor pot experiments using the cotton variety Xinluzao 53 and subjected it to six nitrogen treatments. The data sources for our analysis included hyperspectral and digital images of the cotton leaves. To enhance representation learning capabilities, we enriched the multi-class base learners within each layer of the deep forest (DF) model and introduced skip connections. These enhancements improved the quality of inversion for both hyperspectral and digital image datasets. We then developed image-spectral fusion models, which combined the DF structure with stacking ensemble learning. Our focus was on three levels of fusion: feature-level fusion, decision-level fusion, and secondary decision-level fusion. This approach aimed to further enhance the accuracy and stability of nitrogen content inversion. The DF model satisfied the training condition for small samples. Compared to traditional ML algorithms and the original DF algorithm, the improved DF model achieved an increase in validation set R2 of 13.4–28.5% and 10.9–14.9%, respectively. These findings highlight the enhanced accuracy and stability of the improved DF model. Additionally, compared to the optimal inversion model using two single data sources, the “Image-Spectral” three-level fusion models exhibited improvements in validation set R2 of 8.6–9.3%, 10.5–11.2%, and 11.8–12.5% for feature-level, decision-level, and secondary decision-level fusion, respectively. The improved DF and three-level fusion model collectively contributed to the increased accuracy of cotton nitrogen content inversion. Among these models, the secondary decision-level fusion model demonstrated the most marked improvement. This methodology provides valuable insights into monitoring crop phenotypic parameters in situations with limited sample sizes.

Arctic Sea ice leads detected using sentinel-1B SAR image and their responses to atmosphere circulation and sea ice dynamics

Arctic sea ice leads are linear fractures in pack ice, which provide narrow windows for the enhanced exchange of mass, energy, and momentum between the atmosphere and ocean. However, the parameterisations of the sub-grid (< 1 km) lead distribution and its associated dynamic processes are still challenging for numerical sea ice modelling. This study explores the dynamics governing lead formation in the Arctic Ocean using multiple data sources including Sentinel-1 synthetic aperture radar (SAR) images, buoy array, and atmospheric reanalysis. Random forest (RF) models trained based on the polarisation and texture features of SAR images were compared, and an optimal RF model sensitive to narrow leads was selected with implementation of screening based on lead geometry. As a case study, the lead distribution along the trajectory of buoy array in the central Arctic Ocean during the freezing period of 2018–2019 was obtained. An enhanced lead opening was observed as the sea ice motion swerved from clockwise circulation following the Beaufort Gyre (BG) to meridional advection when assembled into the Transpolar Drift (TPD). Synoptically, active lead-opening events associated with ice divergence were driven by episodic cyclones. We noticed a dramatic change in the backscatter signal over the leads caused by the growth of thin ice and formation of frost flowers. The identification and the evolution of narrow leads during the freezing season given in this study are conducive to increasing our understanding of the response mechanism of sea ice to atmospheric dynamic processes and supporting the numerical simulation of leads.

To prompt or not to prompt: Navigating the use of Large Language Models for integrating and modeling heterogeneous data

Manually integrating data of diverse formats and languages is vital to many artificial intelligence applications. However, the task itself remains challenging and time-consuming. This paper highlights the potential of Large Language Models (LLMs) to streamline data extraction and resolution processes. Our approach aims to address the ongoing challenge of integrating heterogeneous data sources, encouraging advancements in the field of data engineering. Applied on the specific use case of learning disorders in higher education, our research demonstrates LLMs’ capability to effectively extract data from unstructured sources. It is then further highlighted that LLMs can enhance data integration by providing the ability to resolve entities originating from multiple data sources. Crucially, the paper underscores the necessity of preliminary data modeling decisions to ensure the success of such technological applications. By merging human expertise with LLM-driven automation, this study advocates for the further exploration of semi-autonomous data engineering pipelines.

Semantic Web Practices: Infrastructural Politics and the Future of the Web

Semantic Web Practices: Infrastructural Politics and the Future of the Web

Machine learning insights into HIV outbreak predictions in Sub-Saharan Africa

Predicting and preventing HIV outbreaks in Sub-Saharan Africa, a region disproportionately affected by the epidemic remains a significant challenge. This review explores the effectiveness and challenges of using machine learning (ML) for forecasting HIV spread in high-risk areas. ML models have shown promise in identifying patterns and trends in HIV data, enabling more accurate predictions and targeted interventions. ML insights into HIV outbreak predictions leverage various data sources, including demographic, epidemiological, and behavioural data. By analysing these data, ML algorithms can identify high-risk populations and geographical areas susceptible to HIV transmission. This information is crucial for public health authorities to allocate resources efficiently and implement preventive measures effectively. Despite the potential benefits, several challenges exist in using ML for HIV outbreak predictions. These include data quality issues, such as incomplete or inaccurate data, which can affect the reliability of predictions. Additionally, the complexity of HIV transmission dynamics and the need for real-time data pose challenges for ML models. To address these challenges, researchers and practitioners are exploring innovative approaches, such as integrating multiple data sources and using advanced ML techniques. Collaborations between researchers, public health officials, and technology experts are also crucial for developing robust ML models for HIV outbreak predictions. In conclusion, while ML offers valuable insights into HIV outbreak predictions in Sub-Saharan Africa, addressing challenges such as data quality and model complexity is essential for its effective use. By overcoming these challenges, ML has the potential to significantly improve HIV prevention efforts and ultimately reduce the burden of the epidemic in the region. Keywords:   Machine Learning, AI, HIV Outbreaks: Predictions, Insights.

Exploratory Study of Sustainability Practices in Worldwide Major Aircraft Maintenance, Repair, and Overhaul Companies

While sustainable aviation research has primarily focused on the airline industry, aircraft manufacturing, and airport responsibilities, the aircraft maintenance, repair, and overhaul (MRO) industry’s contribution to achieving sustainable aviation goals has been overlooked. This study examines the approaches to corporate sustainability practices used by aircraft MRO companies worldwide. The methodology used to investigate the sustainability actions of major MRO companies globally was mainly based on observational data gathered from multiple data sources, such as environmental, social, and governance (ESG) and investor relations documents. Four companies out of the 13 major MRO companies whose sustainability reports were found are examined as a case study analysis. Each MRO company adopted different definitions and practices of sustainability. The findings show that the aircraft MRO sustainability definition varies among the major companies studied, but some commonalities can be used to develop a proposed definition. The authors propose that aircraft MRO sustainability be defined as the ability to conduct MRO business that positively affects all stakeholders while emphasizing safety, quality, and integrity values, using technology and innovation to make growth, and constantly stabilizing and/or diminishing the impact on the environment. This study contributes to the understanding of sustainable aviation by highlighting the role of MRO companies in achieving sustainable goals.

A STUDY ON IMPROVING SUPPLY CHAIN USING BUSINESS INTELLIGENCE CONCERNING DHARA LOGISTICS

Business intelligence (BI) is software that ingests business data and presents it in user-friendly views such as reports, dashboards, charts and graphs. BI tools enable business users to access different types of data — historical and current, third-party and in-house, as well as semi-structured data and unstructured data like social media. Users can analyze this information to gain insights into how the business is performing. According to CIO magazine: “Although business intelligence does not tell business users what to do or what will happen if they take a certain course, neither is BI only about generating reports. Rather, BI offers a way for people to examine data to understand trends and derive insights.” Organizations can use the insights gained from business intelligence and data analysis to improve business decisions, identify problems or issues, spot market trends, and find new revenue or business opportunities. BI platforms traditionally rely on data warehouses for their baseline information. A data warehouse aggregates data from multiple data sources into one central system to support business analytics and reporting. Business intelligence software queries the warehouse and presents the results to the user in the form of reports, charts and maps. Data warehouses can include an online analytical processing (OLAP) engine to support multidimensional queries. “OLAP provides powerful technology for data discovery, facilitating business intelligence, complex analytic calculations and predictive analytics,” says IBM offering manager Doug Dailey in his data warehousing blog. “One of the main benefits of OLAP is the consistency of information and calculations it uses to drive data to improve product quality, customer interactions and process improvements.” Some newer business intelligence solutions can extract and ingest raw data directly using technology such as Hadoop, but data warehouses are still the data source of choice in many cases.

A Scoping Review of Studies Using Artificial Intelligence Identifying Optimal Practice Patterns for Inpatients With Type 2 Diabetes That Lead to Positive Healthcare Outcomes

The objective of this scoping review was to survey the literature on the use of AI/ML applications in analyzing inpatient EHR data to identify bundles of care (groupings of interventions). If evidence suggested AI/ML models could determine bundles, the review aimed to explore whether implementing these interventions as bundles reduced practice pattern variance and positively impacted patient care outcomes for inpatients with T2DM. Six databases were searched for articles published from January 1, 2000, to January 1, 2024. Nine studies met criteria and were summarized by aims, outcome measures, clinical or practice implications, AI/ML model types, study variables, and AI/ML model outcomes. A variety of AI/ML models were used. Multiple data sources were leveraged to train the models, resulting in varying impacts on practice patterns and outcomes. Studies included aims across 4 thematic areas to address: therapeutic patterns of care, analysis of treatment pathways and their constraints, dashboard development for clinical decision support, and medication optimization and prescription pattern mining. Multiple disparate data sources (i.e., prescription payment data) were leveraged outside of those traditionally available within EHR databases. Notably missing was the use of holistic multidisciplinary data (i.e., nursing and ancillary) to train AI/ML models. AI/ML can assist in identifying the appropriateness of specific interventions to manage diabetic care and support adherence to efficacious treatment pathways if the appropriate data are incorporated into AI/ML design. Additional data sources beyond the EHR are needed to provide more complete data to develop AI/ML models that effectively discern meaningful clinical patterns. Further study is needed to better address nursing care using AI/ML to support effective inpatient diabetes management.

Digital Twin Approach Integrates Existing Facilities, Proposed Developments

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 215074, “Digital Twin for Development and Subsurface Integration: A Window Into the Future,” by Thomas K.G. Gan, Elizabeth Sumadh, and Bhaminie Sagram, Shell. The paper has not been peer reviewed. _ The digital twin (DT) approach provides a virtual and digital representation of existing assets, spanning everything from onshore facilities to offshore platforms, pipelines, and subsurface data, all in one location. It has been used for the operator’s flagship Manatee project to develop three stacked reservoirs offshore Trinidad and Tobago. The DT was used to identify design modifications required at this onshore site and visualize proposed offshore assets. It integrates existing facilities and new design data, including site survey, subsurface, and well data to optimize well planning. Introduction The Manatee field lies within the East Coast Marine Area of Trinidad in approximately 92 m of water and consists of stacked reservoirs of clean and thin-bed unconsolidated sands. Production from these wells is expected to be processed through a new offshore platform and transported by the new pipeline to the Beachfield onshore facility. For a feasibility study and concept-selection stages, the DT platform played a crucial factor in visualizing and contextualizing development options and allowed front-end engineering contractors and operator teams to collaborate across multiple countries and inspect the location virtually. All engineering drawings, maps, design documents, and 3D models were powered by the DT platform. Key Challenges Traditionally, digital collaboration is performed through data and model export from different technical applications and results presented in slides and reporting documents. For oil and gas capital projects, typically 3–6 years elapse from feasibility stage to execution, deepening the issues caused by faulty corporate memory retention, staff movement, and project handover from different internal organizations. Manatee project teams were based in various locations in Trinidad, the US, the UK, the Netherlands, and India, thereby increasing the likelihood of error by using multiple sources of data. Furthermore, effective collaboration without a single source of truth of data can be an obstacle. Given that multiple software was used for model building, development-well planning, and satellite imagery, an inability exists to view 2D and 3D imagery of onshore and offshore facilities with the subsea and subsurface in one place. Furthermore, between 2020 and 2022, the COVID-19 pandemic led to imposed restrictions and limitations placed on site visits. In such scenarios, drone images and video data can prove useful to access facilities remotely. These challenges needed to be addressed by access to a centralized platform that collected, stored, and provided virtual representations to ensure continuous monitoring and efficient decision-making.

Estimating distribution and abundance of wide-ranging species with integrated spatial models: Opportunities revealed by the first wolf assessment in south-central Italy.

Estimating demographic parameters for wide-ranging and elusive species living at low density is challenging, especially at the scale of an entire country. To produce wolf distribution and abundance estimates for the whole south-central portion of the Italian wolf population, we developed an integrated spatial model, based on the data collected during a 7-month sampling campaign in 2020-2021. Data collection comprised an extensive survey of wolf presence signs, and an intensive survey in 13 sampling areas, aimed at collecting non-invasive genetic samples (NGS). The model comprised (i) a single-season, multiple data-source, multi-event occupancy model and (ii) a spatially explicit capture-recapture model. The information about species' absence was used to inform local density estimates. We also performed a simulation-based assessment, to estimate the best conditions for optimizing sub-sampling and population modelling in the future. The integrated spatial model estimated that 74.2% of the study area in south-central Italy (95% CIs = 70.5% to 77.9%) was occupied by wolves, for a total extent of the wolf distribution of 108,534 km2 (95% CIs = 103,200 to 114,000). The estimate of total population size for the Apennine wolf population was of 2557 individuals (SD = 171.5; 95% CIs = 2127 to 2844). Simulations suggested that the integrated spatial model was associated with an average tendency to slightly underestimate population size. Also, the main contribution of the integrated approach was to increase precision in the abundance estimates, whereas it did not affect accuracy significantly. In the future, the area subject to NGS should be increased to at least 30%, while at least a similar proportion should be sampled for presence-absence data, to further improve the accuracy of population size estimates and avoid the risk of underestimation. This approach could be applied to other wide-ranging species and in other geographical areas, but specific a priori evaluations of model requirements and expected performance should be made.

Upscaling vascular aboveground biomass and topsoil moisture of subarctic fens from Unoccupied Aerial Vehicles (UAVs) to satellite level

Arctic and subarctic ecosystems are experiencing rapid changes in vegetation composition and productivity due to global warming. Tundra wetlands are especially susceptible to these changes, which may trigger shifts in soil moisture dynamics. It is therefore essential to accurately map plant biomass and topsoil moisture. In this study, we mapped total, wood, and leaf above ground biomass and topsoil moisture in subarctic tundra wetlands located between Norway and Finland by linking models derived from Unoccupied Aerial Vehicles with multiple satellite data sources using the Extreme Gradient Boosting algorithm. The most accurate predictions for topsoil moisture (R2 = 0.73) used a set of red edge-based vegetation indices with a spatial resolution of 20 m per pixel. On the contrary, wood biomass showed the lowest accuracies across all tested models (R2 = 0.38). We found a trade-off between the spatial resolution and the performance of upscaling models, where smaller pixel sizes generally led to lower accuracies. However, we were able to compensate for reduced accuracy at smaller pixel sizes using Copernicus phenology metrics. A modelling uncertainty assessment revealed that the uncertainty of predictions increased with decreasing pixel sizes and increasing number of co-predictors. Our approach could improve efforts to map and monitor changes in vegetation at regional to pan-Arctic scales.

In search of the ratio of miRNA expression as robust biomarkers for constructing stable diagnostic models among multi-center data.

MicroRNAs (miRNAs) are promising biomarkers for the early detection of disease, and many miRNA-based diagnostic models have been constructed to distinguish patients and healthy individuals. To thoroughly utilize the miRNA-profiling data across different sequencing platforms or multiple centers, the models accounting the batch effects were demanded for the generalization of medical application. We conducted transcription factor (TF)-mediated miRNA-miRNA interaction network analysis and adopted the within-sample expression ratios of miRNA pairs as predictive markers. The ratio of the expression values between each miRNA pair turned out to be stable across multiple data sources. A genetic algorithm-based classifier was constructed to quantify risk scores of the probability of disease and discriminate disease states from normal states in discovery, with a validation dataset for COVID-19, renal cell carcinoma, and lung adenocarcinoma. The predictive models based on the expression ratio of interacting miRNA pairs demonstrated good performances in the discovery and validation datasets, and the classifier may be used accurately for the early detection of disease.

A more complete measure of vertical integration between physicians and hospitals.

To develop an accurate and reproducible measure of vertical integration between physicians and hospitals (defined as hospital or health system employment of physicians), which can be used to assess the impact of integration on healthcare quality and spending. We use multiple data sources including from the Internal Revenue Service, the Centers for Medicare and Medicaid Services, and others to determine the Tax Identification Numbers (TINs) that hospitals and physicians use to bill Medicare for services, and link physician billing TINs to hospital-related TINs. We developed a new measure of vertical integration, based on the TINs that hospitals and physicians use to bill Medicare, using a broad set of sources for hospital-related TINs. We considered physicians as hospital-employed if they bill Medicare primarily or exclusively using hospital-related TINs. We assessed integration status for all physicians who billed Medicare from 1999 to 2019. We compared this measure with others used in the existing literature. We conducted a simulation study which highlights the importance of accurately identifying integrated physicians when study the effects of integration. We extracted physician and hospital-related TINs from multiple sources, emphasizing specificity (a small proportion of nonintegrated physicians identified as integrated). We identified 12,269 hospital-related TINs, used for billing by 546,775 physicians. We estimate that the percentage of integrated physicians rose from 19% in 1999 to 43% in 2019. Our approach identifies many additional physician practices as integrated; a simpler TIN measure, comparable with prior work, identifies only 30% (3877) of the TINs we identify. A service location measure, used in prior work, has both many false positives and false negatives. We developed a new measure of hospital-physician integration. This measure is reproducible and identifies many additional physician practices as integrated.

Personalized Recommendation of College Students' Innovative Entrepreneurship Education Course Content Based on Bayesian Classifier

Computer-aided technology has revolutionized product communication and promotion strategies across industries. With the advent of advanced software tools and digital platforms, businesses now have unprecedented capabilities to create immersive and interactive experiences that effectively showcase their products to consumers. From 3D modeling and rendering software to virtual and augmented reality applications, computer-aided technology enables the creation of highly realistic visualizations and simulations, allowing consumers to interact with products virtually before making purchasing decisions. the integration of Multi-Modal Feature Fusion Classification (MMFFC) computer-aided technology heralds a new era of innovation and effectiveness. MMFFC harnesses the power of multiple data sources and modalities, such as text, images, and videos, to provide a comprehensive understanding of products and consumer preferences. By fusing these diverse features through advanced machine learning algorithms, MMFFC enables businesses to create highly targeted and personalized promotional campaigns. Through sentiment analysis of textual descriptions, image recognition of product attributes, and video analytics of consumer interactions, MMFFC facilitates the generation of rich, multimedia content that resonates with customers on a deeper level. Moreover, MMFFC allows for real-time adaptation and optimization of promotional strategies based on feedback and engagement metrics, ensuring continuous improvement and relevance in an ever-evolving market landscape. The MMFFC computer-aided technology, businesses can enhance their product communication efforts, forge stronger connections with consumers, and drive sales growth in an increasingly competitive marketplace. MMFFC has demonstrated an increase in engagement rates by 20% and a conversion rate improvement of 15%, highlighting its effectiveness in elevating product communication and promotion strategies.

Distributed Denial of Service Attack Detection in IoT Networks using Deep Learning and Feature Fusion: A Review

The explosive growth of Internet of Things (IoT) devices has led to escalating threats from distributed denial of service (DDoS) attacks. Moreover, the scale and heterogeneity of IoT environments pose unique security challenges, and intelligent solutions tailored for the IoT are needed to defend critical infrastructure. The deep learning technique shows great promise because automatic feature learning capabilities are well suited for the complex and high-dimensional data of IoT systems. Additionally, feature fusion approaches have gained traction in enhancing the performance of deep learning models by combining complementary feature sets extracted from multiple data sources. This paper aims to provide a comprehensive literature review focused specifically on deep learning techniques and feature fusion for DDoS attack detection in IoT networks. Studies employing diverse deep learning models and feature fusion techniques are analysed, highlighting key trends and developments in this crucial domain. This review provides several significant contributions, including an overview of various types of DDoS attacks, a comparison of existing surveys, and a thorough examination of recent applications of deep learning and feature fusion for detecting DDoS attacks in IoT networks. Importantly, it highlights the current challenges and limitations of these deep learning techniques based on the literature surveyed. This review concludes by suggesting promising areas for further research to enhance deep learning security solutions, which are specifically tailored to safeguarding the fast-growing IoT infrastructure against DDoS attacks.