Alberto Alfonso Aguilar Lasserre

Pipeline transportation systems (PTS) move oil and gas reliably and economically, but internal and external corrosion threaten their mechanical integrity. This systematic review examines how microscopic analysis, image processing, and artificial intelligence (AI) have been used to identify, monitor, and predict corrosion in oil and gas PTS. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement, we searched Scopus and Web of Science for English- and Spanish-language records from 1987 to 2025, with the last search in August 2026. Of 1085 records, we removed 270 duplicates and excluded 536 that failed three quality-assessment questions. This left 279 studies from 1987 to 2024 for narrative synthesis. We did not apply a formal risk-of-bias tool. AI-based modeling was the largest cluster with 152 studies, followed by corrosion processes (57), microscopy (40), and image processing (30). A total of 141 studies provided open data. Monte Carlo simulation and Bayesian networks were the most commonly applied AI methods to real pipelines, appearing in 82% and 72% of studies, respectively. A sixteen-criterion matrix applied to twenty representative works showed that they met 3 to 12 criteria, with a mean of 5.85. Only two studies covered all four clusters. This result was robust to cluster-weighted scoring (Spearman ρ = 0.94). Integrated, AI-supported corrosion assessment remains uncommon. Limitations include using only two databases, consensus-based screening, and no meta-analysis. The review was not registered and received no external funding.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline, neuroinflammation, and complex alterations involving genetic, metabolic, and environmental factors. In recent years, short-chain fatty acids (SCFAs), derived from gut microbiota fermentation, have attracted particular interest for their potential role in neuroprotective mechanisms and in modulating cognitive decline. In this study, an interpretable computational system, based on Fuzzy Logic (FL) and Artificial Neural Networks (ANN), was developed to analyze patterns of neuroprotection and behavioral loss in a murine model of AD under different dietary and microbial conditions. The Intelligent Decision Support System (IDSS) was built on previously reported experimental evidence from wild-type and transgenic mice fed a control diet, fiber supplementation, or fiber combined with antibiotics. A Mamdani-type fuzzy system was designed to estimate the level of neuroprotection based on acetate, propionate, butyrate, genotype, and diet-related variables, while an artificial neural network was trained to predict behavioral decline based on standard cognitive tests, including the T-Maze, novel object recognition, elevated cross maze, and water maze. The system's purpose was not to replace biological evidence, but rather to organize, simulate, and validate whether the same experimentally observed trend could be reproduced computationally. The results showed that fiber-associated conditions were associated with more favorable neuroprotective profiles, whereas antibiotic-associated conditions were associated with a loss of this benefit. Overall, the proposed system constitutes an interpretable, biologically meaningful tool to support preclinical research on AD and to validate neuroprotective patterns associated with the gut microbiota.

Artificial intelligence (AI) is transforming preclinical research in Alzheimer’s disease (AD), particularly through its integration with Murine experimental models (MEMs) for early detection, monitoring, and therapeutic evaluation. However, current research remains fragmented, with limited standardization and insufficient integration across methodological and translational domains. This systematic literature review (SLR) analyzes 221 peer-reviewed studies published between 2009 and 2025, focusing on AI applications in preclinical AD models. The study takes a closer look at a wide range of AI techniques, from machine learning and deep learning to computer vision, natural language processing, and generative models, mapping out how each is used at various stages of AD and within different experimental approaches. In examining these studies, the results reveal a strong predominance of unimodal approaches, limited use of multimodal data fusion, and significant variability in validation protocols, which collectively hinder reproducibility and translational impact. Additionally, the analysis identifies key gaps in real-time monitoring, explainability, and integration of heterogeneous datasets. Based on these findings, this work proposes a conceptual framework for AI-assisted preclinical pipelines that integrates multimodal data, adaptive learning systems, and translational validation strategies. This framework is designed to enhance potential implications, such as the clinical relevance of preclinical AD research, to facilitate the development of scalable, precision medicine-oriented diagnostic and therapeutic solutions.

Aged pipeline networks face interacting operational, electrochemical, structural, and metallurgical degradation mechanisms that are often assessed independently, limiting diagnosis of system-wide failure propagation. This study develops an integrated, interpreTable fuzzy-logic framework to detect, quantify, and isolate failure scenarios in a 572 km pipeline network. The methodology combines 596 historical SCADA and SAP records, cathodic-protection and soil measurements from 907 evaluation points, 4651 ultrasonic-inspection anomalies, mechanical and metallographic characterization of API L X65 steel, and validated expert knowledge. Four Mamdani fuzzy inference systems were constructed to represent operational capacity, cathodic-protection performance, mechanical integrity, and metallurgical degradation. Their outputs were combined into a unified fault-signature matrix comprising 28 failure scenarios. Scenario FS-3 produced the broadest systemic response by activating all diagnostic residuals. Metallurgical scenarios FS-22 to FS-28 activated 61% of the matrix, showing their extensive influence on pipeline integrity. Structural scenarios FS-15 to FS-21 showed progressively broader signatures as wall deterioration increased. Meanwhile, FS-1, FS-2, FS-5, FS-6, and FS-7 remained localized and were more readily isolated. The proposed framework preserves diagnostic traceability through explicit input variables, fuzzy rules, and residual signatures. It provides an interpreTable basis for failure classification, diagnostic signature breadth, maintenance prioritization, and operator decision support. Its conclusions are limited to the analyzed network and require external validation before application to other pipeline systems.

Pipeline transportation systems (PTS) are widely used maritime and land-based means of transport across several industries, including oil and gas, valued for being reliable, economical, and fast. However, operational instability from multiple interacting variables directly affects the mechanical integrity (MI) of these systems through internal and external corrosion (C). Addressing this problem increasingly relies on microscopic analysis (MA), image processing (IP), and artificial intelligence (AI). Together, these support a horizontal approach to reducing uncertainty and subjectivity in identifying, controlling, and monitoring corrosion-related risk factors. Following the PRISMA 2020 guidelines, this work presents a systematic literature review (SLR) of corrosion-related publications in oil and gas PTS spanning 1987–2025. From 1,085 records identified across two databases, 279 studies were retained after duplicate removal and quality assessment. Of these, 141 provide openly available data. A weighted evaluation matrix scoring twenty representative works against sixteen criteria across four thematic clusters—corrosion type, microscopy, image processing, and AI—shows that the number of jointly satisfied criteria ranges from 3 to 12 (mean 5.85 of 16). This indicates that integrated, multi-technique, AI-supported corrosion assessment remains uncommon across three decades of literature. Beyond consolidating the state of the art into a reusable evaluation instrument, this review establishes the theoretical and bibliometric foundation for an applied, multi-part research line that operationalizes several of these clusters on a real pipeline case study.

Aged pipeline networks are exposed to interacting operational, electrochemical, structural, and metallurgical degradation mechanisms that are commonly assessed independently, limiting the diagnosis of system-wide failure propagation. This study develops an integrated and interpretable fuzzy-logic framework for detecting, quantifying, and isolating failure scenarios in a 572 km pipeline transportation network. The methodology combines approximately 2.5 million historical SCADA and SAP records, cathodic-protection and soil measurements from 89 evaluation points, 5661 ultrasonic-inspection anomalies, mechanical and metallographic characterization of API L X65 steel, and validated expert knowledge. Four Mamdani fuzzy inference systems were constructed to represent operational capacity, cathodic-protection performance, mechanical integrity, and metallurgical degradation. Their outputs were integrated into a unified fault-signature matrix comprising 28 failure scenarios. Scenario FS-3 produced the broadest systemic response by activating all diagnostic residuals, whereas metallurgical scenarios FS-22 – FS-28 activated 61% of the matrix, indicating their extensive influence on pipeline integrity. Structural scenarios FS-15 – FS-21 exhibited progressively broader signatures as wall deterioration increased, while FS-1, FS-2, FS-5, FS-6, and FS-7 remained localized and were more readily isolated. The proposed framework preserves diagnostic traceability through explicit input variables, fuzzy rules, and residual signatures, providing an interpretable basis for failure classification, Diagnostic signature breadth, maintenance prioritization, and operator decision support. Its conclusions are limited to the analyzed network and require external validation before application to other pipeline systems.

Climate change has had profound impacts on agricultural systems, altering crop productivity, changing precipitation patterns, spreading pests and diseases, reducing soil quality, displacing agricultural areas, and increasing the use of inputs such as fertilizers and pesticides, which in turn leads to an increase in atmospheric emissions. To address this issue, this research proposes the use of a multi-agent system-based model to analyze the vulnerability of sugarcane production, representing complex systems and adapting to changing conditions by integrating dynamic and uncertain variables. The main advantage of the model is that it enables the quantification and analysis of critical variables, including the use of fuel, fertilizers, and nitrogen oxide (N₂O) emissions. The results demonstrate how the increase in operating trend negatively impacts environmental performance, highlighting the fragility of the system. Meanwhile, the validation of the model through structural tests and extreme conditions confirmed its reliability in supporting decision-making processes. Likewise, the average vulnerability value of the system (0.54) indicates a moderately unstable condition, susceptible to climatic and economic changes. Complementarily, the IMPACT 2002+ methodology was applied to conduct a life cycle assessment (LCA) of sugarcane, encompassing its cultivation and industrial processing. It was found that the resources used in sugar mills have the most significant environmental impact in the categories of climate change, human health, ecosystem quality, and resource consumption. This impact is caused by CO₂ emissions, the use of toxic pesticides and heavy metals, and high dependence on fossil fuels such as coal, natural gas, and oil, mainly. These findings underscore the need to enhance environmental management in Mexico's sugar sector by adopting cleaner technologies, establishing reliable ecological databases, and implementing assessment tools such as multi-agent modeling and life cycle analysis.