Explore the vast range of titles available.

Background Intra-hospital transport of critically ill, mechanically ventilated patients is associated with significant physiological risks. Adaptive Minute Ventilation (AMV) is a closed-loop ventilation mode that automatically maintains target minute ventilation. While AMV has shown promise in intensive care unit settings, its performance during intrahospital transport has not yet been evaluated. This trial aims to assess the safety and efficacy of AMV compared with conventional ventilation strategies during patient transport. Methods This is a single-centre, open-label, randomized controlled non-inferiority trial conducted at Peking Union Medical College Hospital. Fifty adult patients receiving invasive mechanical ventilation and requiring transport will be randomly assigned (1:1) to either AMV or their pre-transport ventilation mode. The primary endpoint is the change in arterial carbon dioxide pressure (PaCO₂) from 30 min before to 10 min after transport. Secondary outcomes include changes in oxygenation index (PaO₂/FiO₂), peak airway pressure, dynamic compliance, and rapid shallow breathing index. Randomization will be performed using permuted block randomization. Both intention-to-treat and per-protocol analyses will be conducted. Missing data will be handled using multiple imputation. The study protocol has been reviewed and approved by the Ethics Committee of Peking Union Medical College Hospital (Approval No. I-25PJ0781). It was registered in the Chinese Clinical Trial Registry (ChiCTR2500109659) on 23 September 2025 as a retrospective registration. Discussion This trial will be the first randomized study to evaluate AMV during intrahospital transport of critically ill patients. The findings will address an important evidence gap in transport ventilation strategies, potentially informing clinical practice on whether AMV provides a safe and effective alternative to conventional modes. By employing a non-inferiority design and standardized transport protocols, the study seeks to generate robust and clinically relevant evidence despite its single-centre setting and relatively small sample size. Trial registration Approved by the Ethics Committee of Peking Union Medical College Hospital (Approval No. I-25PJ0781). Trial registration: ChiCTR, ChiCTR2500109659. Registered 23 September 2025, https://www.chictr.org.cn/hvshowproject.html?id=284847 .

Background It is well known that lipids abnormally accumulate in the alveoli during idiopathic pulmonary alveolar proteinosis (PAP). It is unclear, however, whether lipids also abnormally accumulate in serum. This study investigated the serum lipid panels in idiopathic PAP patients and explored the relationships between serum levels and the severity of idiopathic PAP. Methods and Results Clinical data including the level of serum lipids were evaluated in 33 non-diabetic idiopathic PAP patients and 157 healthy volunteers. Serum levels of triglyceride were higher in PAP patients than in healthy subjects (median: 192.00 mg/dl ( P 25 : 104.36, P 75 : 219.00) vs 119.56 mg/dl ( P 25 : 78.81, P 75 : 193.03), P < 0.05), while high-density lipoprotein cholesterol (HDL-C) levels were lower in patients than in the control group (42.50 ± 10.30 vs 51.34 ± 12.06 mg/dl, P < 0.01). Forced expiratory volume in one second and forced vital capacity in hypertriglyceridemia patients were lower than those in patients with normal triglyceride. Serum LDL-C and HDL-C ratio correlated negatively with PaO 2 (r = -0.403, P < 0.05) and positively with lactate dehydrogenase (r = 0.381, P < 0.05). Conclusions PAP associates with high triglyceride and low HDL levels in the serum, and these lipids provide potential intervention strategy for treatment.

Recent large audio language models (LALMs) demonstrate remarkable capabilities in processing extended multi-modal sequences, yet incur high inference costs. Token compression is an effective method that directly reduces redundant tokens in the sequence. Existing compression methods usually assume that all attention heads in LALMs contribute equally to various audio tasks and calculate token importance by averaging scores across all heads. However, our analysis demonstrates that attention heads exhibit distinct behaviors across diverse audio domains. We further reveal that only a sparse subset of attention heads actively responds to audio, with completely different performance when handling semantic and acoustic tasks. In light of this observation, we propose HeadRouter, a head-importance-aware token pruning method that perceives the varying importance of attention heads in different audio tasks to maximize the retention of crucial tokens. HeadRouter is training-free and can be applied to various LALMs. Extensive experiments on the AudioMarathon and MMAU-Pro benchmarks demonstrate that HeadRouter achieves state-of-the-art compression performance, exceeding the baseline model even when retaining 70% of the audio tokens and achieving 101.8% and 103.0% of the vanilla average on Qwen2.5-Omni-3B and Qwen2.5-Omni-7B, respectively.

In recent years, long short-term memory (LSTM) has been successfully used to model sequential data of variable length. However, LSTM can still experience difficulty in capturing long-term dependencies. In this work, we tried to alleviate this problem by introducing a dynamic skip connection, which can learn to directly connect two dependent words. Since there is no dependency information in the training data, we propose a novel reinforcement learning-based method to model the dependency relationship and connect dependent words. The proposed model computes the recurrent transition functions based on the skip connections, which provides a dynamic skipping advantage over RNNs that always tackle entire sentences sequentially. Our experimental results on three natural language processing tasks demonstrate that the proposed method can achieve better performance than existing methods. In the number prediction experiment, the proposed model outperformed LSTM with respect to accuracy by nearly 20%.