Acetaminophen (APAP) overdose can cause severe liver injury and life-threatening conditions that may lead to multiple organ failure without proper treatment. N-acetylcysteine (NAC) is the accepted and prescribed treatment for detoxification in cases of APAP overdose. Nonetheless, in acute liver failure (ALF), particularly when the ingestion is substantial, NAC may not fully restore liver function. NAC administration in ALF has limitations and potential adverse effects, including nausea, vomiting, diarrhoea, flatus, gastroesophageal reflux, and anaphylactoid reactions. Mesenchymal stromal cell (MSC)-based therapies using paracrine activity show promise for treating ALF, with preclinical studies demonstrating improvement. Recently, MSC-derived extracellular vesicles (EVs) have emerged as a new therapeutic option for liver injury. MSC-derived EVs can contain various therapeutic cargos depending on the cell of origin, participate in physiological processes, and respond to abnormalities. However, most therapeutic EVs lack a distinct orientation upon entering the body, resulting in a lack of targeting specificity. Therefore, enhancing the precision of natural EV delivery systems is urgently needed. Thus, we developed an advanced targeting technique to deliver modified EVs within the body. Our strategy aims to employ bioorthogonal click chemistry to attach a targeting molecule to the surface of small extracellular vesicles (sEVs), creating exogenous chimeric antigen receptor-modified sEVs (CAR-sEVs) for the treatment. First, we engineered azido-modified sEVs (N-sEVs) through metabolic glycoengineering by treating MSCs with the azide-containing monosaccharide N-azidoacetyl-mannosamine (Ac4ManNAz). Next, we conjugated N-sEVs with a dibenzocyclooctyne (DBCO)-tagged single-chain variable fragment (DBCO-scFv) that targets the asialoglycoprotein receptor (ASGR1), thus producing CAR-sEVs for precise liver targeting. The efficacy of CAR-sEV therapy in ALF models by targeting ASGR1 was validated. MSC-derived CAR-sEVs reduced serum liver enzymes, mitigated liver damage, and promoted hepatocyte proliferation in APAP-induced injury. Overall, CAR-sEVs exhibited enhanced hepatocyte specificity and efficacy in ameliorating liver injury, highlighting the significant advancements achievable with cell-free targeted therapy.

In the rapidly evolving healthcare landscape, artificial intelligence (AI), particularly the large language models (LLMs), like OpenAI's Chat Generative Pretrained Transformer (ChatGPT), has shown transformative potential in emergency medicine and critical care. This review article highlights the advancement and applications of ChatGPT, from diagnostic assistance to clinical documentation and patient communication, demonstrating its ability to perform comparably to human professionals in medical examinations. ChatGPT could assist clinical decision-making and medication selection in critical care, showcasing its potential to optimize patient care management. However, integrating LLMs into healthcare raises legal, ethical, and privacy concerns, including data protection and the necessity for informed consent. Finally, we addressed the challenges related to the accuracy of LLMs, such as the risk of providing incorrect medical advice. These concerns underscore the importance of ongoing research and regulation to ensure their ethical and practical use in healthcare.

Objective: To investigate the association between patient-to-physician ratios, a measure of physician workload, and various patient outcomes in the emergency department (ED). Methods: This retrospective observational study analyzed 406,220 ED visits at a tertiary care center in Taipei, Taiwan, between January 2015 and December 2019. The dynamic patient-to-physician ratio was calculated using minute-by-minute data to reflect real-time physician workload. Multivariable regression models, adjusted for potential confounders, assessed the association between this ratio and 7-day mortality (primary outcome), ED length of stay, waiting time, and medical expenses (secondary outcomes). Generalized additive models were used to explore non-linear relationships. Sensitivity analyses evaluated alternative mortality timeframes, missing data handling, and a simplified patient-to-physician ratio. External validation was performed using data from two additional hospitals. Results: Higher patient-to-physician ratios were significantly associated with increased odds of 7-day mortality. Compared to ratios of less than 10, the adjusted odds ratios were 1.46 (95% CI 1.16, 1.83) for ratios between 10 and 19, 1.79 (95% CI 1.43, 2.25) for ratios between 20 and 29, and 1.95 (95% CI 1.53, 2.49) for ratios of 30 and above. Similar trends of increased risk were observed for longer ED length of stay, prolonged waiting times, and higher medical expenses. Sensitivity analyses using alternative mortality timeframes, missing data handling methods, and the simplified patient-to-physician ratio yielded consistent results, supporting the main findings. Conclusions: Higher patient-to-physician ratios are associated with worse outcomes for ED patients. Our findings suggest that maintaining ratios below 10 may be ideal for optimizing care quality, while ratios exceeding 20 pose significant risks to patients.

The early detection of cardiac arrest (CA) in emergency departments (EDs) is crucial for patient safety. However, existing deep-learning research often neglects irregular time intervals between measurements and the challenge of performance degradation in short sequences. The limited accessibility of medical data further complicates the external validation of models. To address these issues, we developed a deep learning-based early warning system accommodating variable-length and irregularly measured time series data. Our system includes three models: A Time Mask Temporal Convolutional Network (TM-TCN) incorporates a missing value mask to address the problem of missing values in multivariate time series, and univariate time series with time intervals are used to ensure that the model can detect the rapid deterioration of patients. Finally, we use a designed fusion method to enable the system to make better predictions for short sequence samples. Our system achieved an area under the receiver operating characteristic curve (AUROC) of 0.9831 and an area under the precision-recall curve (AUPRC) of 0.2150 in the experiment of 8 h before CA on the National Taiwan University Hospital dataset. In the external validation, the proposed system achieved an AUROC of 0.9734 and an AUPRC of 0.1336 8 h before CA on the Far Eastern Memorial Hospital dataset and obtained an AUROC of 0.8428 and an AUPRC of 0.0533 0 to 8 h before CA on the MIMIC-IV-ED dataset. These results demonstrate the system’s reliability and adaptability across datasets, highlighting its potential to advance healthcare informatics research by addressing critical challenges in time series data modeling.

Pei-Lin Lee serves as Clinical Associate Professor, School of Medicine, National Taiwan University; Consultant, Center of Sleep Disorder, National Taiwan University Hospital. Her current academic positions at international sleep societies include American Academy of Sleep Medicine Fellow and Co-Chair International Assembly; Asian Society of Sleep Medicine, Sleep Medicine Education Task Force committee member. Her current research focuses on the era of new technology and big data in sleep medicine; and intervention on sleep and metabolism in sleep disordered breathing.