Abstract Purpose Large language models (LLMs) have proven performance for certain diagnostic tasks; however, limited studies have evaluated their consistency in recommending appropriate medication regimens for a given diagnosis. Medication management is a complex task that requires synthesis of drug formulation and complete order instructions for safe use. Here, the performance of GPT-4o, an LLM available with OpenAI’s ChatGPT, was tested on 3 medication management tasks. Methods GPT-4o performance was tested on 3 medication tasks: identifying available formulations for a given generic drug name, identifying drug-drug interactions (DDIs) for a given medication regimen, and preparing a medication order for a given generic drug name. For each experiment, the model’s raw text response was captured exactly as returned and evaluated using clinician evaluation in addition to standard LLM metrics, including Term Frequency-Inverse Document Frequency (TF-IDF) vectors, normalized Levenshtein similarity, and Recall-Oriented Understudy for Gisting Evaluation (ROUGE-1/ROUGE-L) F1 score between each response and its reference string. Results For the first task of drug-formulation matching, GPT-4o had 49% accuracy for generic medications being matched to all available formulations, with an average of 1.23 omissions per medication and 1.14 hallucinations per medication. For the second task of drug-drug interaction identification, the accuracy was 54.7% for identifying the DDI pair. For the third task, GPT-4o generated order sentences containing no medication or abbreviation errors in 65.8% of the cases. Conclusion Model performance for basic medication tasks was consistently poor. This evaluation highlights the need for domain-specific training through clinician-annotated datasets and a comprehensive evaluation framework for benchmarking performance.
An explainable transformer model learning from entire treatment timelines for pan-cancer risk profiling across healthcare systems
Abstract Cancer outcomes vary widely between individual patients, each accumulating an irregular record of treatments, diagnoses, measurements, and complications. Current prognostic models reduce this complexity into a single snapshot, focus on narrow clinical settings, and rarely generalize across hospitals. Here we introduce Chronicle , an explainable transformer that learns from entire patient trajectories to predict diverse clinical outcomes throughout the disease course while capturing both short- and long-term temporal dependencies. Trained on 53.7 million longitudinal data points from 51,711 patients spanning 67 cancer types, Chronicle operates natively on irregular data without imputation and jointly predicts eight endpoints within a flexible framework adaptable to additional outcomes. Chronicle outperformed cross-sectional models for overall survival prediction (C-index 0.84 vs 0.76-0.79), stratified patients more accurately than established prognostic systems, including TNM stage, and predicted seven adverse event and transfusion endpoints (AUC 0.80-0.92). Applied without retraining to 69,341 patients in Germany, Switzerland, and the United States, Chronicle generalized across healthcare systems and improved further with local fine-tuning. Integrated explainability traced each risk update to patient-specific clinical factors, revealing distinct temporal persistence of prognostic information, with relevance half-lives ranging from weeks for therapies to nearly one year for baseline characteristics. These findings demonstrate that learning from hospital-wide patient trajectories enables interpretable and continuously updated predictions, providing a scalable framework to support individualized treatment decisions.
From Opinions to Outcomes: Optimizing Survey Methods and Principles Applicable to the Delphi Process in Critical Care Research
BACKGROUND. Survey research in critical care has dramatically increased over the past 25 years, and use of surveys within Delphi methods, which provide consensus definitions for heterogeneous disease states, has similarly expanded significantly. It is crucial for survey researchers to understand the nuances of survey development and expectations of information provided by surveys, to use them correctly. Few simple sources exist to guide development considerations to generate high-quality survey research. No studies to our knowledge exist that draw parallels between the principles of survey research as a guide within the Delphi process, as one potential conduit between a steering committee and expert panel. METHODS. We present a narrative review of and framework for rigorous design of survey research, highlighting the applicability to the design of surveys administered within Delphi methods. RESULTS AND CONCLUSIONS. The methodologic considerations presented for survey research and Delphi methods can inform high-quality critical care research.
Simulations and Brain Death: A Novel Approach to Evaluating Competency (P7-15.011)
Objectives: Consensus clinical criteria for refractory septic shock were recently developed, but the incidence and outcomes with these criteria are unknown. There were two aims: 1) to describe the incidence of refractory septic shock and 2) to assess outcomes of patients with septic shock meeting refractory septic shock criteria compared with those who did not. Design: Multicohort, retrospective study. For aim 1, a descriptive epidemiologic study was designed to ascertain the cumulative incidence of refractory septic shock in patients with septic shock. For aim 2, a cohort study was conducted to compare hospital mortality between patients meeting refractory septic shock clinical criteria compared with those who did not. Setting: Large U.S. health system from 2012 to 2024. Patients: Adults (age ≥ 18 yr) meeting both Centers for Disease Control and Prevention Adult Sepsis Event and Sepsis-3 clinical criteria for septic shock. Patients with missing hospital discharge disposition and incomplete vasopressor dosage information were excluded. Interventions: None. Measurements and Main Results: Clinical criteria for refractory septic shock were operationalized as concomitant occurrence of norepinephrine (base) equivalent greater than 0.5 µg/kg/min and lactate greater than 2 mmol/L. Risk-adjusted hospital mortality was assessed with a generalized linear mixed-effects model with a random effect for hospital and fixed effects for age, sex, Elixhauser Comorbidity Index, and Sequential Organ Failure Assessment (original description, without cardiovascular component). Of 35,914 patients assessed, 15,732 patients with septic shock were included. Refractory septic shock occurred in 3423 patients (21.8%), with a cumulative incidence of 218 cases (95% CI, 211–224) per 1000 patients with septic shock. Hospital mortality with refractory septic shock was 64.4%, and risk-adjusted hospital mortality odds were 4.87-fold higher (95% CI, 4.46–5.31) compared with patients without refractory septic shock. Conclusions: Clinical criteria for refractory septic shock were fulfilled in about one in five patients with septic shock and associated with substantially higher mortality.
Decompression with or without Duraplasty for Chiari I and Syringomyelia
A38-09 Secondhand Aerosol From a Non-heated, Non-combustion (NHNC) Nicotine Nebulization Device vs Heated Ends and Heated-tobacco: A 6-subject Crossover Study
Abstract Rationale Secondhand exposure to electronic nicotine products is driven by exhaled emissions. Eliminating heat and combustion may reduce exhaled toxic carbonyls. Methods In a 4-day within-subject crossover, six adults each used four devices after 24-hour abstinence: an NHNC nebulization device (Bluesky), a coil-heated ENDS (RELX), a heated-tobacco device (IQOS), and a heated aerosol device (Marlboro). For each condition, paired pre-use exhaled blank and post-use exhaled samples were collected; 50 L of exhaled breath was captured post-use. Nicotine and 12 carbonyls were quantified (DNPH/activated-carbon with HPLC). Primary endpoint: within-subject change (Δ, post-pre) in total carbonyls (μg/50 L). Secondary: Δ acetaldehyde and Δ acrolein. Results The exhaled carbonyls(within-subject change, post-pre) in total differed markedly across devices: Bluesky (non-heated, non-combustion; NHNC) 0.85 ± 6.99, RELX 4.41 ± 2.58, IQOS 62.38 ± 21.77, Marlboro 80.73 ± 20.24 (p = 0.0015). Bluesky had significantly lower total exhaled carbonyls than IQOS (p = 0.0077) and Marlboro (p = 0.0021, μg/50 L; mean±SD), with a nonsignificant trend versus RELX (p = 0.945). For acetaldehyde, Bluesky was lower than all three (Bluesky 0.20 ± 0.16 vs. RELX 1.93 ± 0.86, p = 0.016; vs IQOS 43.08 ± 16.70, p = 0.0044; vs Marlboro 65.40 ± 13.69, p = 0.00025, μg/50 L). For acrolein, average exhaled volume was 0.54 ± 6.94 (Bluesky), 1.35 ± 2.05 (RELX), 7.94 ± 6.07 (IQOS), and −1.86 ± 5.09 (Marlboro); pairwise differences vs Bluesky were not significant after correction (p ≥ 0.37, μg/50 L). Conclusions In a controlled, within-subject comparison using exhaled breath as a proxy for secondhand aerosol, the NHNC nebulization device (Bluesky) produced near-zero average increment in total exhaled carbonyls, whereas heated devices—particularly heated-tobacco—showed order-of-magnitude higher increases. These descriptive data support the hypothesis that removing heat and combustion from nicotine aerosolization can materially reduce exhaled carbonyl burden relevant to secondhand exposure. Larger, randomized studies with standardized capture and enhanced nicotine quantification are warranted. This abstract is funded by: Pneuma Respiratory.Inc
C56-02 Endotracheal Intubation Simulation “Mock Intubation” in Clinical Practice: The Role of In Situ Simulation
Abstract Introduction Emergent endotracheal intubation in the intensive care unit (ICU) is a high-risk, time-sensitive procedure performed to maintain airway patency and ensure adequate oxygenation and ventilation, often complicated by hypotension, hypoxemia, or cardiac arrest. Successful outcomes depend on precise team coordination, communication, and situational awareness. Drawing inspiration from the demonstrated benefits of simulation-based cardiopulmonary resuscitation training in enhancing team performance, this study examines the feasibility of implementing in situ “mock intubation” in the ICU. The objective is to determine whether this approach can be seamlessly integrated into routine practice to improve teamwork, procedural readiness, and patient safety. Methodology A prospective interventional study was conducted in the medical ICU at Mayo Clinic Rochester over two months. ICU staff—including physicians, advanced practice providers (APP), nurses, and respiratory therapists—participated in eight scheduled 45-minute in situ mock intubation sessions integrated into routine clinical hours. Each session included five participants (one physician faculty, one physician trainee/APP, two nurses, and one respiratory therapist) and incorporated case-based scenarios requiring completion of the intubation and management of post-procedural complications. Structured debriefings followed each session to identify performance gaps, enhance team communication, and assess the educational value and feasibility of incorporating mock intubation into routine ICU practice. Results A total of eight mock intubations were conducted with 40 participants. Sessions were seamlessly integrated into the routine clinical workflow without disruption. Participants reported high satisfaction, unanimously agreeing that the sessions were valuable and should be continued. Feedback emphasized psychological safety and realism: “This offered a safe space to communicate,” one noted; another stated, “It’s important to practice where it’s easy to make mistakes,” and a third remarked, “Real-life experiences were brought into simulations.” These themes underscore participants’ desire for ongoing sessions and their perceived real-world relevance. Conclusion Mock intubation using in situ simulation provides a practical approach to training for emergent endotracheal intubation, enhancing team communication, coordination, and situational awareness in the ICU. This on-site format eliminates the need for off-unit scheduling and was well received by staff. The potential of in situ mock intubation to mirror the success of simulated cardiopulmonary resuscitation training highlights the importance of team-based procedural preparation in high-acuity environments. The intervention fosters procedural readiness, teamwork, and a culture of safety. Given its feasibility, clinical relevance, and positive reception, in situ mock intubation should be integrated into routine ICU practice as a team training tool for high-risk procedures. This abstract is funded by: None
Pancreatic tumor microenvironment reprogramming via alloantigen-expressing virotherapy elicits tumor rejection and improves immunotherapy response
Background Pancreatic ductal adenocarcinoma (PDAC), the most common malignant type of pancreatic cancer, is characterized by a dense desmoplastic stroma, low neoantigen burden, and a highly immunosuppressive tumor microenvironment (TME). Methods We present a novel strategy harnessing acute transplant rejection mechanisms by employing a recombinant oncolytic rVMG vector engineered to express murine alloantigens H-2Kb (rVMG-H-2Kb) or H-2Kk (rVMG-H-2Kk), thereby inducing tumor-specific antigenic mismatch responses. Results In vitro, rVMG-H-2Kb and rVMG-H-2Kk exhibited strong replication and cytolytic activity while inducing cell surface expression of both H-2Kk and endogenous H-2Kb, in addition to upregulation of antigen presentation genes (β2-microglobulin, Tap1, and Tapbp). In two immunocompetent PDAC models, intratumoral and systemic delivery of rVMG-H-2Kb and rVMG-H-2Kk delayed tumor progression, with rVMG-H-2Kk conferring a survival advantage. Multiplex immunohistochemistry and immunophenotyping revealed substantial TME remodeling, marked by increased effector T-cell infiltration, regulatory T-cell depletion, and reduced fibrosis. Spatial transcriptomics further showed compartment-specific immune activation and epithelial metabolic reprogramming, corroborating with enhanced tumor immunogenicity. Despite these effects, rVMG-H-2Kk also induced compensatory immunosuppressive pathways, including upregulation of antiviral response genes and immune checkpoint receptors such as programmed death-ligand-2. Importantly, combination therapy with rVMG-H-2Kk and murine checkpoint blockade (anti-programmed cell death protein-1 and anti-cytotoxic T-lymphocyte-associated protein 4) drastically improved survival compared to checkpoint blockade alone. Strikingly, surviving mice resisted tumor rechallenge, indicating the establishment of durable antitumor memory. Conclusion These findings establish rVMG-mediated alloantigen delivery as a novel immunotherapeutic platform capable of converting immune-cold tumors into immune-hot, sensitizing tumors to immune checkpoint inhibitors, and establishing durable antitumor immunity in PDAC.