Bio-Bras

The growing demand for more personalized, functional, and sensorially appealing foods has driven the adoption of computational approaches in the food sector. In this context, the application of artificial intelligence to sensory science emerges as a promising alternative to automate evaluations, predict consumer preferences, and enhance innovation in food products. The aim of this study was to map the international scientific output on the topic through a bibliometric analysis of 735 articles indexed in the Web of Science database between 1991 and 2025. The Bibliometrix (R) and VOSviewer software tools were used to extract indicators of productivity, impact, collaboration, and thematic structure. The results show a significant growth in publications since 2016, with a strong concentration in the last 5 years. China, India, and Brazil rank among the most productive countries, with notable contributions from authors based in Australia and East Asia. The co-occurrence analysis revealed seven thematic clusters, highlighting approaches based on machine learning, neural networks, computer vision, and electronic sensors. This study contributes to the understanding of the field’s evolution and provides insights to inform research and development strategies in the food sector.

Accurate electricity demand forecasting is crucial for the efficient operation of Brazil's Southeast/Center-West power subsystem, the largest component of the national grid. This study investigates the impact of incorporating exogenous variables in hourly demand forecasting using three artificial intelligence models: Random Forest, Multi-Layer Perceptron, and Long Short-Term Memory networks. The models were evaluated across three distinct temporal scenarios: pre-pandemic (2019), pandemic (2021), and postpandemic (2023). Results demonstrate that incorporating exogenous variables significantly improves forecasting accuracy under stable conditions but yields limited benefits during major structural disruptions, such as the COVID-19 pandemic. These findings highlight the importance of context-aware model selection strategies and provide valuable insights for power system operators in designing adaptive forecasting frameworks for modern grid management.

This article presents a systematic review of cybersecurity governance, analyzing current practices and the main challenges organizations face in managing cyber risks. It highlights the critical role of governance structures that involve not only IT, but also senior management, to protect both the organization and its digital value chain. The review uses the Automatic Literature Mapping Selection method, supported by machine-learning-based decision trees and complementary metrics such as the Scientific Significance Score (SSS) and the Time-Adjusted Publication Score (TAPS), which prioritize journal quality, scientific impact, and recency. The results expose persistent gaps. A key issue is the weak integration between corporate strategy and cybersecurity initiatives, which undermines the implementation of effective security measures. Another problem is the lack of clear, standardized metrics to assess cybersecurity effectiveness, as many organizations still rely on superficial indicators, often limited to the number of reported incidents. The review also finds a scarcity of empirical studies evaluating the return on cybersecurity investments and the real effectiveness of adopted tools. Finally, the rapid evolution of technologies such as the Internet of Things (IoT) and Artificial Intelligence (AI) creates new security challenges that demand more adaptive governance frameworks; however, the literature still offers few practical approaches capable of addressing the specific risks introduced by these innovations.

Background: Fluid resuscitation is essential for perioperative management in patients undergoing major surgery. The spleen contributes to cardiovascular adjustments during volume loading through its reservoir function. The splenic artery resistance index (SARI) reflects splanchnic perfusion and vascular resistance, and postoperative changes may indicate altered hemodynamic states, including bleeding and hemoglobin decline. Objectives: This study aimed to determine the correlation between the splenic artery resistance index (SARI), measured by Doppler ultrasonography, and postoperative hemoglobin decline, the number of packed red blood cell (PRBC) units transfused, and the bleeding severity score. Design: Observational study. Setting: Intensive Care Unit (ICU) of the University Hospital of Brasília. Patients: A total of 41 adult patients admitted to the ICU after surgeries lasting more than 4 hours under general anesthesia. Interventions: None. Measurements and Main Results: Simultaneous evaluations were performed within the first 24 hours after ICU admission and repeated 48 hours later, including serum lactate, standard hemodynamic parameters, Sequential Organ Failure Assessment (SOFA) score, central venous oxygen saturation (ScvO₂), hemoglobin concentration, and the number of packed red blood cell (PRBC) units transfused. At ICU admission, the splenic artery resistance index (SARI) showed a negative correlation with hemoglobin and a positive correlation with the number of PRBC units transfused. ScvO₂ correlated negatively with the need for transfusion. For predicting significant blood transfusion (≥2 units), SARI showed an area under the curve (AUC) of 0.99, with a cutoff value of ≥0.66, sensitivity of 97.5%, and specificity of 100%, whereas ScvO₂ showed an AUC of 0.48 with a cutoff value of ≤65%, sensitivity of 55%, and specificity of 95%. At ICU admission, SARI also showed a strong positive correlation with serum lactate levels, and patients without lactate clearance ≥20% had a higher frequency of elevated SARI (≥0.66) at 48 hours (85% vs. 18%). Conclusion: The Splenic Artery Resistance Index (SARI) correlated with hemoglobin levels and transfusion needs, showing better accuracy than conventional parameters in identifying postoperative bleeding risk within 48 hours after ICU admission.

This study introduces a novel zero-shot methodology for detecting center pivot irrigation systems (CPIS) using Sentinel-1 radar time series and the Segment Anything Model (SAM). The proposed pipeline addresses key challenges in remote sensing by combining a time series image composition, a sliding window approach, post-segmentation refinements, and domain-specific geometric heuristics to identify CPIS accurately with SAM. The methodology includes a custom RGB color composite derived from three spaced time steps in the radar time series, enhancing the visualization of temporal patterns for CPIS detection. A sliding window mechanism with multi-pass refinements ensures complete spatial coverage and resolves boundary inconsistencies. Overlapping masks are filtered using a custom non-maximum suppression (NMS) algorithm, sorted by area. Furthermore, a heuristic approach based on semicircle fitting classifies CPIS into circular, semicircular, and pizza-slice geometries while filtering irregular shapes using a normalized standard deviation metric. Experimental results demonstrate the method’s ability to segment CPIS, achieving more than 95% per-object accuracy with minimal false positives and negatives, eliminating the need for annotated datasets or model retraining.

This study introduces a novel iterative active learning framework for semantic segmentation of remote sensing images, designed to minimize annotation effort while achieving competitive performance. The framework consists of an integrated interface that allows point labeling and model training in a single platform, along with the proposition of a new loss function, the Dynamic Weighted Confidence Dice Loss, which dynamically penalizes confident incorrect points. To highlight the method’s efficiency even under strict annotations, we labeled at most one foreground and one background pixel per frame in each iteration. The experiments were conducted on the BSB Aerial dataset, containing RGB bands with 0.24m image spatial resolution. The dataset included 1000 training, 200 validation, and 200 test frames, each 256×256 pixels, in a binary scenario (road and background). The framework was compared against dense annotations, random sparse annotations, and the standard Dice loss (with the same selected points in the iterative process). Dense annotations, which required labeling all pixels, achieved an Intersection over Union (IoU) of 89.63%, serving as an upper bound, while random sparse annotations reached an IoU of 75.20%, serving as a lower bound. By the sixth iteration, our method achieved an IoU of 86.46%, only 3.17% lower than dense annotations, while labeling just 0.0136% of the total dataset. Furthermore, the proposed loss consistently outperformed the Dice loss during iterative training, with IoU improvements ranging from +0.17% in the first iteration to +4.17% in the final iteration, highlighting that its effectiveness increases as more points in error-prone regions are annotated.

Objective With colorectal cancer (CRC) being a leading cause of cancer-related deaths globally, early detection through screening is critical for improving survival rates. Recent guidelines recommend lowering the initial screening age from 50 to 45 years based on increasing CRC incidence among younger adults. Methodology This retrospective study evaluates the impact of this adjustment on adenoma detection rates (ADR), a validated quality indicator of screening colonoscopies. We conducted a retrospective analysis of 2,792 patients who underwent colonoscopy screening at a major healthcare institution. Patients were categorized into three age groups: 45-49, 50-59, and 60-75 years. We compared ADR across these groups with a focus on evaluating the consistency of ADR in the newly included younger age group against established benchmarks. Result The ADRs for the 45-49 age group were 51.2% (105/205) in male patients and 38.5% (116/301) in female patients, closely mirroring those of the 50-59 age group at 51.08% (188/368) in male patients and 34.80% (133/382) in female patients, with no significant reduction in detection rates. Gender-specific analysis revealed higher ADR in male patients across all age groups. These findings were statistically significant when comparing procedure type with age group and gender (p < 0.05). Conclusion The study supports maintaining the already established ADR for the new age group of 45-49 as the ADR was comparable to those in the older groups. Similar polyp biopsy +/- removal rates in the 45-49-year-old age group (43.6%; 221/506) compared to the older age group (42.7%; 320/750) is also an indirect measure of the effectiveness of early screening for colon cancer.