Lung ultrasound (LUS) is a bedside tool for assessing pulmonary edema in patients at risk due to heart failure or impaired kidney function. However, automated LUS analysis remains challenging because of speckle noise, imaging artifacts, and operator-dependent acquisition variability. In this work, we present a deep learning framework for multi-class LUS video classification that explores two components: hierarchy-aware training, and anatomy-guided learning. Starting from a strong baseline, we introduce hierarchical training strategies and then introduce pleural line mask supervision to guide model attention toward anatomically relevant regions. We study four clinically relevant classes--healthy, B-lines, consolidations, and mixed B-lines with consolidations--using an open-access dataset of 1,886 videos from 219 patients, evaluated with patient-level five-fold cross-validation. Results show that hierarchy-aware training improves pathological separation relative to flat classification, while mask-guided attention supervision achieves the highest mean macro-F1 of 65.7\% and produces more localized attention patterns. Transfer experiments on the external COVID-BLUeS dataset further show competitive and parameter-efficient adaptation while preserving pleural-focused attention behavior. These findings suggest that combining clinically structured objectives with anatomy-guided supervision is a practical approach to robust, interpretable LUS video analysis. Code and model implementations are available at https://github.com/Alya-Almsouti/LUS-video-classification.
Alya Almsouti, Lotfi Abdelkrim Mecharbat, Noha Aboukhater et al.· arXiv.org· 0 citations
Sepsis is a time-critical condition associated with substantial morbidity and mortality, where delays in recognition and treatment markedly worsen outcomes. Although machine learning models show promise for early detection, their clinical impact has been constrained by poor integration into workflows, limited interpretability, and insufficient support for coordinated action. This study introduces a human-centered, systems-based agentic AI architecture for sepsis risk modeling and proactive clinical management. Rather than generating static risk scores, the system continuously interprets evolving patient data, situates risk within the clinical workflow, and supports timely, clinician-supervised interventions. Grounded in systems engineering and guided by the Systems Engineering Initiative for Patient Safety (SEIPS) framework, the architecture embeds predictive intelligence within the broader socio-technical work system, enabling closed-loop monitoring, coordination of safety-critical tasks, and feedback-driven adaptation. By reframing sepsis prediction as an adaptive, workflow-aware safety intervention, this approach advances AI from passive decision support toward an accountable, action-oriented partner in care delivery while preserving clinician oversight.
A patient-focused, systems-based framework for AI-enhanced VR training in home PD, informed by human-centereddesign is presented, which integrates realistic procedural simulations with AI-driven feedback on sequencing and sterile technique, while modeling the complete PD workflow within the home as a safety-critical care environment.
Sarah Ahmed Alkindi, Saed Amer, Mecit Can Emre Simsekler et al.· AHFE International· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
