Deepfake technology has moved well beyond research laboratories, with synthetic video and audio now being deployed to commit corporate fraud, impersonate executives, and manipulate employees at scale. Most detection systems in the research literature fail in three recurring ways: they perform well in controlled tests but poorly in real deployment; they handle only one modality at a time; and they require computing resources that most organisations do not possess. This research addresses these problems through a detection framework built on a fine-tuned EfficientNetV2-S visual backbone combined with three independent Radial Basis Function (RBF) Support Vector Machine (SVM) classifiers, covering visual, audio, and combined audio–visual detection, with the classification stage adding 2.3 ms per frame to a pipeline that requires 195 ms per frame on a single CPU thread. Evaluation followed the official FaceForensics++ source-video split, with all partitioning performed at video level. On FaceForensics++ C23, the visual classifier achieved an AUC-ROC (Area Under the Receiver Operating Characteristic Curve) of 0.9848 (95% CI 0.9787 to 0.9898) at the frame level and 0.9937 at the video level, with an accuracy of 95.53%. On a held-out DeepFakeDetection subset comprising unseen-manipulation methods and subjects, AUC-ROC was 0.9689. Cross-dataset evaluation on Celeb-DF v2 without retraining produced an AUC-ROC of 0.6589, a drop of 32.59 percentage points, with 65.6% of real frames flagged as fake at the inherited operating threshold; a recalibration experiment shows that no operating point on that curve is usable. On the full ASVspoof 2019 Logical Access evaluation partition, which contains thirteen attack types absent from training, the audio classifier reached an equal error rate of 16.55% (95% CI 14.70 to 18.40), ranging from 0.36% on the most detectable attack to 33.43% on the least. A controlled ablation on FakeAVCeleb v1.2, using identity-disjoint partitions and separate labels for each modality, showed a visual-only AUC-ROC of 1.0000 and audio-only of 0.9884. Score-level fusion reduced AUC-ROC to 0.9776 (DeLong p = 0.016), because averaging discards information when video and audio are manipulated independently. A governance framework is presented that maps each technical limitation to a specific organisational control, grounded in proportionality, independence, and adaptability and aligned to ISO/IEC 27001:2022 Annex A, the National Institute of Standards and Technology (NIST) Cybersecurity Framework 2.0 and the NIST AI Risk Management Framework. The detection results establish that this governance layer cannot depend on detector output once the deployment domain differs from the training domain.
High-cost medical equipment procurement creates long-term clinical, technical, and service dependencies, making the pre-tender stage a critical decision point. A multicriteria decision analysis proof of concept for computed tomography procurement was developed using six macrocriteria and three hypothetical scenarios. Of 43 submissions, 42 were eligible, 40 had complete Analytic Hierarchy Process matrices, and 23 met the consistency ratio threshold of 0.10 for primary weighting. Scenario-specific weights were elicited for hypothetical hospital settings: S1, a proximity hospital; S2, a high-demand general hospital; and S3, a differentiated referral hospital. These weights were applied to five hypothetical profiles using SAW, TOPSIS, and MOORA, with agreement and sensitivity analyses. Reliability and service continuity ranked highest in S1; clinical fit and reliability were nearly tied in S2; and strategic value and interoperability led S3. The balanced profile led all methods in S1 and S2, whereas TOPSIS selected a strategic profile in S3. Under one-at-a-time ±20% weight changes, first place persisted in 36/36, 36/36, and 33/36 classifications for S1, S2, and S3. In S3, a 7.1% C6 increase triggered the nearest switch, supporting reporting stability alongside rank. Findings support methodological feasibility under controlled computed tomography scenarios; application to other equipment or clinical domains requires context-specific adaptation and prospective institutional validation with real data.
Children exposed to violence require a comprehensive, multidisciplinary support network operating within a well-coordinated child protection system. The effectiveness of such systems depends fundamentally on strong intersectoral collaboration and high-quality relationships among professionals from different sectors. Enhancing system performance therefore requires systematic identification and mitigation of management barriers that hinder coordinated action. This study proposes a Multi-Criteria Decision Analysis (MCDA) model to identify, analyze, and prioritize management barriers within child protection networks through an integrated MCDA framework for complex systems. The Fuzzy Delphi Method (FDM) was used to screen the most relevant barriers, swing weighting was used to elicit the relative importance of the retained barriers, the Analytic Hierarchy Process (AHP), applied in its absolute measurement mode, was used to derive priorities for the predefined performance levels and support the sector-specific evaluation of three candidate decision alternatives, and Social Network Analysis (SNA) degree centrality was used to weight the judgments of the four participating experts (one per network sector) in the final aggregation. The proposed approach was implemented in the Child Protection Network of a medium-sized Brazilian municipality. The results show that the Unified Information System alternative achieved the highest overall priority (w = 0.3525), closely followed by the Creation of an Intersectoral Management Group (0.3450) and Continuous Training for Network Professionals (0.3025); given the narrow margin (approximately 0.0075) between the two top-ranked alternatives, this prioritization should be interpreted as an indicative rather than a definitive ranking. More broadly, the results demonstrate that (i) the proposed framework was successfully applied, providing a structured and transparent process for identifying and prioritizing critical management barriers, and (ii) the resulting prioritization supports decision-makers and practitioners in allocating resources more effectively, strengthening intersectoral coordination, and improving services for children and adolescents exposed to violence.
Teacher well-being has expanded rapidly as a research field, yet conceptual and methodological heterogeneity continues to limit cumulative knowledge. This study examined the intellectual and thematic development of teacher well-being research, how teacher well-being is measured in influential studies, and which theoretical models underpin its conceptualization. We conducted a systematic review complemented by science-mapping analyses of Web of Science and Scopus records published between 2015 and 2026. The full corpus comprised 1113 articles. Historiography, thematic evolution, and trend topics were examined with Bibliometrix. The top citation-impact decile (111 articles, based on annualized citations) was screened for in-depth analysis of measurement and theoretical models, yielding a final high-impact analytical subset of 92 empirical studies and 178 applications of instruments or indicators. The historiographic analysis identified multiple partially parallel intellectual trajectories, while more recent indexed publications showed greater representation of resilience, psychological well-being, retention, and contextual approaches alongside stress, depression, and burnout. Within this 92-study high-impact subset, measurement remained fragmented: 46.6% of applications assessed ill-being, 64.6% used non-teacher-specific measures, and no positive teacher well-being instrument was dominant. Within the same subset, although 82.6% of studies had explicit theoretical grounding, no model accounted for one quarter of the sample; JD-R was most frequent (19.6%), followed by Positive Psychology/PERMA (17.4%). Teacher well-being was primarily modeled as an outcome (63.0%). Teacher well-being research appears more diversified than consolidated. Its advancement depends less on adopting a single theory or instrument than on strengthening alignment between construct definition, theoretical function, and measurement, while promoting clearer measurement standards and context-sensitive psychometric evidence.
Forest and funnel plots are the standard graphical outputs of evidence synthesis, yet the most widely used R functions draw them directly on the graphics device, which makes them hard to restyle, combine with other figures, or embed in a reproducible workflow. We present ggmeta, a ggplot2 extension that builds forest and funnel plots from meta package objects or from tidy data frames and returns them as ordinary ggplot objects, with each plot component implemented as a separate ggproto layer. The package handles twenty-one summary measures with measure-specific back-transformation, can pool effects from a data frame by the inverse-variance and DerSimonian–Laird methods, handles subgroups and table columns, and provides style presets modelled on JAMA, BMJ, and RevMan 5. We demonstrate ggmeta on five open datasets from published meta-analyses covering risk ratios, odds ratios, mean differences, Fisher-z correlations, and logit proportions. Plotted summaries matched the source objects exactly, the standalone pooling routine agreed with metafor and meta to machine precision, and the drawn coordinates matched the input values. A benchmark against five other R forest-plot tools reports code length, rendering time, and memory use. All code and data are distributed under the GPL (≥3) licence.
