Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 3 appraisals
PloS one
Risk assessment of civil aviation cabin safety incidents based on the CNN-LSTM-Attention model
Against the backdrop of the thriving global civil aviation industry, cabin safety, which is a crucial link in civil aviation safety systems, has prompted an increasingly urgent need for intelligent risk management. This study constructs an integrated framework of "deep text mining-intelligent risk assessment" and proposes a hybrid CNN-LSTM-Attention model for cabin safety incident classification. Using 8,280 global cabin abnormal event reports from 2004-2024, the model achieves 95.01% accuracy and an F1 score of 94.17% on the test set, substantially outperforming benchmark approaches. Compared with XGBoost (89.45% accuracy, 88.98% F1), the proposed model improves accuracy by 5.56% and the F1 score by 5.19%, demonstrating superior capability in extracting deep semantic features and identifying key risk patterns. The framework also establishes a "text feature-risk mechanism-hazard level" mapping system, enabling more interpretable risk quantification. The findings provide an effective technical path for intelligent early warning of cabin safety risks and offer methodological support for data-driven decision-making in civil aviation safety management.
30 May 2026
Read appraisal →PloS one
Enhancing aviation safety: An 80-year data-driven model for classification of aviation incident and accident
The aviation system is safety-critical by nature, and any occurrence of an incident or accident can lead to the loss of human life and significant operational disruptions. The International Civil Aviation Organization (ICAO) emphasizes that every flight must take off and land safely-a goal achieved over 126,000 times daily. Despite major advancements,mishaps and accidents continue to occur, underscoring the need for robust safety management systems. The accurate classification of aviation occurrences (Incident or Accident) reports is essential for safety management, yet manual review is time-consuming and prone to inconsistency. While incident/accident labels are assigned during reporting, automated classification enables rapid triage, detection of potential mislabeling, and support for severity assessment in high-volume aviation safety operations. To address this,we developed and compared three machine learning classifiers-Multinomial Naive Bayes, Random Forest, and Support Vector Machine-using TF-IDF vectorization on an 80-year dataset of 53,770 aviation occurrence summaries obtained from the Transportation Safety Board of Canada. A two-stage evaluation strategy was employed, consisting of an initial 80/20 train-test split to create an independent test set, followed by 5-fold cross-validation applied exclusively to the training data to ensure robustness and prevent optimistic bias. The Support Vector Machine (SVM) classifier achieved the highest classification performance, attaining an accuracy of 98.06% during 5-fold cross-validation, with consistent results across folds, demonstrating its effectiveness in managing high-dimensional textual data and dataset complexity. The proposed framework provides a robust foundation for automated aviation safety report processing, offering practical value for (1) early triage of safety reports, (2) identification of potentially mislabeled cases requiring expert review, and (3) integration into downstream severity assessment pipelines. This work advances beyond prior classification studies by establishing a benchmark on the largest historical aviation safety dataset while delivering a deployable and operationally relevant framework for real-world safety management applications. The findings offer valuable insights for regulatory authorities and airline operators, contributing to enhanced safety oversight, improved response strategies, and safer aviation operations.
23 May 2026
Read appraisal →Neuroscience
Neural responses to acute hypoxia and hyperoxia
The brain's high metabolic demands make it highly sensitive to oxygen level changes, with both hypoxia and hyperoxia affecting cognitive and neural function. However, the underlying neural mechanisms remain unclear. This study aimed to characterise the effects of acute normobaric hypoxia and hyperoxia on neural function and perceptual thresholds using EEG, visual evoked potentials (VEPs), and critical flicker fusion (CFF) thresholds. In a single-blind, randomised, crossover design, thirty participants inhaled 10.5%, 21%, or 100% oxygen while resting-state EEG, VEPs, and CFF thresholds were recorded. Hypoxia reduced CFF thresholds in an exposure-dependent manner, while hyperoxia produced a smaller, transient reduction. Resting-state EEG revealed significantly increased power across delta, theta, and beta bands in response to hypoxia. Whereas hyperoxia induced smaller increases in theta and alpha and decreases in gamma power and a reduction in beta power over the sensorimotor area. Hypoxia increased spectral entropy, indicating impaired network synchronisation, while hyperoxia increased fuzzy entropy, particularly in eyes-closed states, potentially reflecting enhanced neural complexity. In VEPs, hypoxia reduced P1 amplitude and increased N2 deflections, reflecting disrupted early and late sensory processing, with sustained effects across blocks. Hyperoxia transiently increased P1 amplitude and showed minor effects on other components. These findings demonstrate the impact of altered cerebral oxygen on fundamental neural processing and the associated state- and time-dependent effects on brain function. This provides insight into the mechanisms underlying oxygen-induced neuromodulation and highlights the value of EEG and CFF as biomarkers for tracking neural responses to oxygen variability.
16 Apr 2026
Read appraisal →