Citations

Abstract Rationale Propofol-based sedation in the intensive care unit (ICU) is associated with clinically significant hemodynamic instability, including hypotension. Previous studies have identified hypotension as a key predictor of adverse outcomes in critically ill patients. However, no validated tools currently exist to systematically predict this risk. Therefore, this study aimed to develop and externally validate machine-learning (ML) models for predicting the risk of hypotension in ICU patients receiving propofol sedation. Methods We developed ML models using retrospective data from adult, mechanically ventilated ICU patients who received propofol infusion for at least 24 hours across four Mayo Clinic sites between 2018 and 2024. The primary outcome was hypotension, defined as ≥ 2 consecutive MAP readings ≤60 mmHg within 15 minutes during the first 6 hours of propofol sedation. The MIMIC-IV dataset was used for external validation. Features were selected using a union of Boruta and LASSO-penalised logistic regression. Four algorithms (elastic net-penalised logistic regression, random forest, LightGBM, and multilayer perceptron) were tuned using Bayesian optimization to minimize cross-entropy loss during leave-one-site-out cross-validation (LOSO-CV). The algorithm with the lowest cross-entropy loss during LOSO-CV was chosen as the best-performing algorithm. Classification thresholds were selected to target sensitivity ≥80%. Model performance was evaluated using area under the receiver operating characteristic curve (AUC-ROC), calibration slope and intercept, sensitivity, and specificity. Feature importance was evaluated using SHapley Additive exPlanations (SHAP). Results The development cohort included 16,361 patients from Mayo Clinic, of whom 3,482 (21.3%) developed hypotension. The external validation cohort comprised 4,684 patients from MIMIC-IV, with 115 (2.5%) experiencing hypotension. Patient characteristics included age, admission Sequential Organ Failure Assessment (SOFA) scores, starting propofol dose, ICU admission source, and comorbidities. The best-performing algorithm was LightGBM. The average AUC-ROC during LOSO-CV was 0.70 (95% confidence interval [CI] 0.68-0.71), versus 0.64 during external validation. The average calibration slope and intercept during LOSO-CV was 1.08 (95% CI 0.89-1.27) and 0.08 (95% CI -0.25-0.41), respectively, versus 0.78 and -2.07 during external validation. The average sensitivity and specificity during LOSO-CV was 0.81 (95% CI 0.79-0.83) and 0.47 (95% CI 0.42-0.52), respectively, versus 0.54 and 0.65 during external validation. Conclusions We developed and externally validated an ML-based classifier that demonstrates moderate discrimination for predicting propofol-induced hypotension in mechanically ventilated ICU patients. This tool may facilitate early identification of high-risk patients and enable proactive hemodynamic management strategies. Further prospective validation and assessment of clinical impact are warranted. This abstract is funded by: None

Abstract Background Ventilated patients in the intensive care unit are often complex and require multiple care team members including bedside nurses, respiratory therapists, advanced practice providers, and physicians. A Quality Improvement initiative was created to address any barriers in communication. Spontaneous breathing trials (SBT) are often used as a surrogate for multidisciplinary communication. Both Respiratory therapist (RT) and provider led spontaneous breathing trial protocols are often performed but not necessarily documented or communicated appropriately. Our primary aim was to increase spontaneous breathing trial documentation from a baseline of 12% to 80% with a counterbalance of not increasing the 24-hour reintubation rate. Methods Using the DMAIC framework, we defined the gap in quality by getting baseline patient data for 3 months including SBT completion rate, SBT documentation rate, time to first SBT, total time on ventilator, 24-hour reintubation rate, and total ICU days. All care team members including providers, RTs, and nurses completed a 10-questions Likert 5-scale survey regarding their views of multidisciplinary communication and SBT completion. We identified the lack of documentation of SBTs and implemented a standardized respiratory therapy shift summary note which included SBT details and major ventilator changes. We then collected the same data over 3 months after implementation. Results Post intervention analysis revealed documentation of daily spontaneous breathing trials improved from 12% to 79%. Pre and post intervention means for patient focused outcomes were statistically unchanged for time to first SBT 26 hours vs 32 hours (p = 0.16), Total ventilator time 5.6 vs 5.4 (p = 0.83), total ICU days 1.9 days vs 2.7 days (p = 0.11), and reintubation rate 14% vs 15%.Survey results showed that there was a statistically significant improvement in overall satisfaction regarding multidisciplinary communication, satisfaction regarding ventilator management discussion on rounds, and if documentation amount was appropriate. Also, there was a significant improvement in the percentage of care team members that were aware of daily SBTs occurring. Conclusion The implementation of RT shift summary greatly improved staff member satisfaction, multidisciplinary discussion during rounds, and awareness of SBT occurrence among providers, RTs, and nurses. While patient centered outcomes were statistically unchanged over this very brief period of observation, having better staff satisfaction and communication may eventually lead to improved outcomes over time and can lead to further analysis to avoid mishaps that can arise from miscommunication. This abstract is funded by: None

Abstract Rationale Hypophosphatemia affects about one-third of intensive care unit (ICU) patients and is linked to longer hospital stays, illness severity and poor outcomes, yet guidelines for monitoring and correction remain unclear. Studies show a strong association between hypophosphatemia and both prolonged invasive mechanical ventilation (IMV) and weaning failure in patients with respiratory failure, due to reduced adenosine triphosphate generation and weakened respiratory muscle performance. Although phosphate abnormalities correlate with higher mortality in sepsis and septic shock, their impact on IMV duration in this population remains unclear. Therefore, we aimed to evaluate the independent association between hypophosphatemia during ICU stay and the duration of IMV in patients with septic shock. Methods We studied adult patients with septic shock admitted to the medical ICU at Mayo Clinic, Rochester, between March 1, 2006, and April 17, 2018. A retrospective analysis was conducted to examine the association between hypophosphatemia—defined as any serum phosphate level < 2.5 mg/dL during the ICU stay—and the duration of IMV in hours. The following linear regression models were used: 1, unadjusted; 2, adjusted for baseline demographics (age, sex, weight); 3, adjusted for baseline respiratory comorbidities and Charlson Comorbidity Index (CCI); and 4, adjusted for illness severity (using APACHE III and SOFA scores) and oxygenation (minimum PaO2/FiO2 score during ICU stay). For each model, the β-coefficient was reported, which represents additional hours of IMV associated with hypophosphatemia, along with 95% confidence intervals and p-values. Results The study included 1082 patients, among them 653 (60.4%) required IMV. 53.6% (n = 350) developed hypophosphatemia during their ICU stay. The mean duration of IMV was significantly longer in patients with hypophosphatemia compared with those without (184 ± 218 h vs 100 ± 126 h; p < 0.001; 95% CI 56-112). This association persisted across all multivariable linear regression models. After adjustment for baseline demographics, hypophosphatemia remained significantly associated with longer IMV duration (β = 79.8 h; 95% CI 51.8-107.8; p < 0.001). Further adjustment for respiratory comorbidities and CCI yielded similar results (β = 81.8 h; 95% CI 54.2-109.4; p < 0.001). After controlling for illness severity and oxygenation status, hypophosphatemia remained independently associated with prolonged IMV duration (β = 69.2 h; 95% CI 39.9-98.4; p < 0.001). Conclusion Hypophosphatemia was independently associated with longer duration of IMV in patients with septic shock, which prolongs ICU level care and delays rehabilitation. This highlights the need for updates on guidelines for maintaining optimal serum phosphate levels in patients who are mechanically ventilated. This abstract is funded by: None

Abstract Introduction Sedation is routinely administered to facilitate comfort and tolerance of life support for critically ill adults dependent on invasive mechanical ventilation. Isoflurane, a widely used anesthetic during surgery, has properties that may also benefit ICU patients requiring continuous sedation. However, efficacy as a sedative and safety in this population is not well established. Methods We conducted two identically designed multicenter randomized trials, INSPIRE ICU1 and INSPIRE ICU2, at hospitals across the US. Ventilator-dependent adults were randomly assigned in a 1.5:1 ratio to receive either inhaled isoflurane or intravenous propofol as the first-line sedative for up to 48 ± 6 hours, targeting a sedation depth between Richmond agitation-sedation scale (RASS) -1 (drowsy) to -4 (responsive only to physical stimuli). Isoflurane was administered via an anesthetic conserving device (Sedaconda ACD-S). In both arms, opioids were permitted at the lowest dose required to achieve freedom from pain. The primary outcome was the percentage of time at target sedation level, assessed as a noninferiority outcome with pre-specified margin of 15%. Secondary outcomes included opioid exposure, wake-up time after end of treatment, time to extubation, and proportion of time with spontaneous breathing effort. Results A total of 470 combined participants were included in the two trials. In each trial, 142 patients were randomly assigned to isoflurane and 93 to propofol. In INSPIRE ICU1, percentage of time within target sedation depth was 77.1 [53.2, 89.9] % with isoflurane and 80.9 [58.2, 96.4] % with propofol (difference -1.5 [-8.9, >99] %; p < 0.01 for noninferiority). In INSPIRE ICU2, percentage of time within target sedation depth was 72.6 [66.4, 78.9] % with isoflurane and 74.5 [67.5, 81.5] % with propofol (difference -1.9 [-9.4, >99] %; p < 0.01 for noninferiority). In both trials, patients assigned to isoflurane received significantly less opioid, measured as fentanyl equivalents, during the treatment period. In INSPIRE ICU2, participants receiving isoflurane were significantly likelier to exhibit spontaneous breathing efforts, but this finding was not replicated in INSPIRE ICU1. There were no significant differences in time to wake-up or time to extubation between groups in either trial. No episodes of malignant hyperthermia or propofol infusion syndrome were reported. Adverse events were comparable in type and frequency between groups. Conclusions In two US multicenter trials, isoflurane was noninferior to propofol for maintaining target sedation depth in adults undergoing invasive ventilation, and was associated with significantly lower opioid exposure. (NCT05327296, NCT05312385) This abstract is funded by: Sedana Medical

Abstract Rationale Propofol sedation in mechanically ventilated patients often causes clinically significant hypotension. While we concurrently developed and validated a machine-learning (ML) classifier to predict hypotension risk, accurately estimating the timing of hemodynamic deterioration could enable more precise, proactive interventions. This study aimed to develop and externally validate ML models that predict not only whether hypotension will occur, but also when it is likely to develop after propofol initiation. Methods ML models were developed using retrospective data from adult, mechanically ventilated ICU patients who received propofol infusion for ≥24 hours across four Mayo Clinic sites from 2018 to 2024 and developed hypotension (≥2 consecutive MAP readings ≤60 mmHg within 15 minutes) during the first 6 hours of sedation. The MIMIC-IV dataset was used for external validation. Features were selected using a union of Boruta and LASSO-penalised quantile regression. Two quantile algorithms (LASSO-penalised quantile regression and LightGBM) were trialed to predict the conditional median and interquartile range (IQR) of time to hypotension. In addition, two probabilistic algorithms (NGBoost and Mixture Density Network) were trialed to predict the full conditional distribution. The quantile models were tuned via Bayesian optimisation to minimise pinball loss during leave-one-site-out cross-validation (LOSO-CV), whereas probabilistic models were tuned to minimise the Continuous Ranked Probability Score. The algorithm with the lowest average pinball loss was deemed the best-performing algorithm. Model performance was evaluated using median absolute error (MAE), IQR coverage, and IQR width. Feature importance was evaluated using SHapley Additive exPlanations (SHAP). Results We included 3482 patients from the Mayo Clinic for model development, with 115 patients from the MIMIC-IV dataset used for external validation. Patient characteristics in the modeling dataset included age, sex, admission SOFA scores, weight, ICU admission source, and initial propofol dose. The best-performing algorithm was LightGBM-based quantile regression. The average MAE during LOSO-CV was 4.11 hours (95% confidence interval [CI] 3.98-4.23), versus 4.83 hours during external validation. The average IQR coverage and width during LOSO-CV was 0.49 (95% CI 0.48-0.49) and 7.12 (6.89-7.35), respectively, versus 0.07 and 6.51 during external validation. Conclusions We developed and externally validated a quantile regression model that predicts both the timing and uncertainty of propofol-induced hypotension in mechanically ventilated ICU patients. This model provides clinicians with actionable time estimates that may facilitate preemptive hemodynamic interventions and optimize sedation management. Prospective validation, expanded predictors and clinical impact assessment are warranted. This abstract is funded by: None

Abstract Rationale Postoperative pulmonary complications (PPCs) importantly contribute to postoperative morbidity and mortality, particularly for American Society of Anesthesiologists (ASA) physical status III patients. It remains unclear whether risk factors for PPCs depend on sex. We hypothesised that sex modifies the relationship between perioperative risk factors and PPCs in ASA III patients undergoing noncardiothoracic surgery with general anesthesia. Methods We performed a secondary analysis of prospective data from 1,202 ASA III patients undergoing non-cardiothoracic surgery. Data consisted of preoperative demographics, comorbidities, intraoperative physiology, medications, and procedure information. The primary outcome was a composite of PPCs by postoperative day 7. Mixed-effects linear models incorporated sex as main effect and interaction with covariates. Results Age, chronic obstructive pulmonary disease, upper abdominal surgery, surgical duration, and emergency surgery significantly increased the odds of ≥ 1 PPC by postoperative day 7. Male patients presented a higher rate of ≥ 3 PPCs, and mechanical ventilation was less protective in female patients. Yet, sex did not significantly modify the relationship between perioperative factors and the primary outcome (no p<0.05 for the interaction effect with sex). In the secondary analysis, gastroesophageal reflux disease affected atelectasis more strongly in male than female patients (relative odds-ratios for female versus male =0.27; 95% CI 0.13 to 0.57). Conclusions In ASA III patients undergoing non-cardiothoracic surgery with general anesthesia, perioperative factors are predominantly not significantly modified by sex in their relationship to PPCs. GERD’s association with atelectasis in male patients could be explored in future prospective studies as a target for reducing pulmonary complications. This abstract is funded by: NHLBI - UH3 HL140177

Abstract Rationale Lung-protective ventilation protocols use predicted body weight (PBW) to set tidal volumes. However, biases exist in these formulas, with overestimation of lung capacity in older, shorter, female, and non-White patients—groups that experience higher mortality rates during invasive mechanical ventilation (IMV). This study investigated associations between pre-admission forced vital capacity (FVC), driving pressure, and patient outcomes. Methods This multicenter retrospective study included adults (18-95 years old) with pre-admission spirometry who required IMV between 01/01/2016 and 05/31/2025 across the Mayo Clinic, spanning 11 ICUs within 5 hospitals in 4 states. Patients were stratified by FVC/PBW ratio quartiles. The primary outcome was 28-day mortality. Trends were assessed using linear or logistic regression, with FVC/PBW as a continuous variable. Mediation analysis examined pathways between FVC/PBW and mortality, with driving pressure (cm H2O, standardized) as the mediator, and adjusted for covariates including age, sex, SOFA score, and BMI. Absolute risk differences and number needed to harm (NNH) were estimated for key thresholds (e.g. FVC/PBW ≥45 mL/kg, PFTs within 2 years). Results Among 4,839 patients (38.7% female, mean (±SD) age 64.5±13.5 years), FVC/PBW ratio demonstrated substantial individual variation, with a mean of 48.3 ± 13.4 mL/kg. There was systematic variation by demographics (Figure): females had significantly lower ratios than males (45.4 versus 50.2 mL/kg, P < 0.001), and ratios were lower with older age from 52.5 mL/kg in patients <65 to 43.2 mL/kg in those ≥85 years (P < 0.001 for trend). Patients in the lowest FVC/PBW quartile had higher 28-day mortality compared to the highest quartile (26.0% versus 17.0%, P < 0.001) and higher driving pressures (14.0 versus 10.0 cm H2O, P < 0.001; Figure). In mediation analysis (n = 4,147), driving pressure mediated the relationship between FVC and mortality. Each 1-SD increase in FVC/PBW was associated with 0.364-SD lower driving pressure (P < 0.001) and 2.4% lower mortality. The adjusted indirect effect through driving pressure was significant (ACME -0.024, 95% CI: -0.029 to -0.019), accounting for the mortality effect. Sensitivity analyses across binary thresholds, age quartiles, outlier exclusion, and obstructive pathology confirmed robust mediation. Patients with FVC/PBW <45 mL/kg (N = 1,644; 39.6%) had 28-day mortality of 17.8% versus 13.5% for ≥45 mL/kg (risk difference 4.3%, 95% CI 1.8-6.8%; NNH: 23 patients). Conclusions Pre-admission lung function was strongly associated with mortality in adults receiving IMV through driving pressure-mediated pathways. Driving pressure is a modifiable target, especially in patients with smaller lung capacities at highest risk for ventilator-induced lung injury. Further prospective investigation is needed to confirm these findings. This abstract is funded by: National Heart, Lung, and Blood Institute (NHLBI) of the National Institute of Health (NIH) Grant Number K23HL151671

Abstract Rationale Community-acquired pneumonia (CAP) remains a leading cause of morbidity and mortality worldwide; current guidelines recommend either the use of β-lactam therapy with a macrolide (azithromycin; AZM) or monotherapy with a fluoroquinolone for treatment. Limited evidence exists on their comparative effectiveness. To address this gap, we examined the differences in outcomes among CAP patients treated with β-lactam therapy plus AZM or fluoroquinolone monotherapymethods: We emulated a randomized clinical trial that hypothetically assigned patients to β-lactam therapy plus AZM or fluoroquinolone. Eligible patients were all adults hospitalized for nonsevere community-acquired pneumonia (CAP) based on ATS/IDSA criteria; at Mayo Clinic hospitals (3 academic and 16 community) between May 2018 and September 2022 and received one of the two treatments within 12 hours of admission. Ensemble machine learning with augmented inverse probability weighting (AIPW) was used to evaluate “intention to treat” effects on hospital-free days (HFDs), 180-day mortality, use of advanced respiratory support, in-hospital mortality, and the composite of advanced respiratory support and/or in-hospital mortality. We used the E-value and post-discharge systolic blood pressure (SBP) (negative outcome) to evaluate effects of potential unmeasured confounding and alternative “doubly robust” inverse probably weighting with regression adjustment (IPWRA) to evaluate robustness of the findings. Results 2,451 patients (2,142 β-lactam therapy plus AZM and 309 fluoroquinolone) were eligible and included in the analyses: 53.9% were female, 86.8% were White, and mean age was 58.4 years (SD: 17.02). Compared with β-lactam therapy plus AZM, the fluoroquinolone group experienced significantly higher HFD (1.50 days; 95% CI: 1.08, 1.91), lower use of advanced respiratory support (RR: 0.32; 95% CI: 0.09, 0.54), lower in-hospital mortality (RR: 0.14; 95% CI: 0.00, 0.40), and composite of advanced respiratory support and/or in-hospital mortality (RR: 0.29; 95% CI: 0.08, 0.49). Moderate to large e-values (ranging from 3.50 to 4.74) and use of negative outcomes (post-discharge SBP) suggested that unmeasured confounding is unlikely to change the findings. Sensitivity analysis using IPWRA showed very similar results. Conclusion Among hospitalized patients with non-severe CAP, treatment with fluoroquinolone monotherapy was associated with more hospital-free days, lower in-hospital mortality, and reduced need for advanced respiratory support compared with AZM and β-lactam combination. The findings highlight the role of fluoroquinolone as a beneficial alternative CAP patients. However, further studies are needed to explore the underlying mechanisms driving these differences and assess whether specific patient subgroups may benefit most from quinolone-based therapy. This abstract is funded by: none

Abstract Rationale In fungal endemic regions, differentiating benign from malignant pulmonary nodules on CT scans remains a major diagnostic challenge. Coccidioides nodules frequently resemble malignant lesions on CT and PET imaging, presenting significant challenges for differentiation, even among radiologists. Consequently, the benign biopsy and resection rate, especially in the Southwestern US, remains high. We previously developed [DOI:10.1371/journal.pone.0196910] and externally validated [DOI:10.1183/13993003.02485-2020] the Benign versus aggRessive nODule Evaluation using Radiomic Stratification (BRODERS) machine learning (ML) classifier to non-invasively distinguish between benign and malignant pulmonary nodules with an AUC of 0.87. However, its performance in certain benign nodules with malignant surface and texture characteristics like in coccidioidomycosis had a suboptimal AUC of 0.65. In this study, we redesigned our radiomics model by including numerous clinical features. Methods In this retrospective study, 113 pulmonary nodules (66 Coccidioides and 47 adenocarcinoma spectrum) detected on CTs with ground truth pathology diagnosis were identified by screening for key terms like “nodule”, “cocci”, and “adenocarcinoma” from both CT and pathology reports using a large language model trained on clinical textual data (Illuminate®). Clinical features such as demographics, smoking history, and laboratory studies were extracted by trained clinicians from the electronic medical record. All nodules were segmented and radiomic features extracted using a semi-automated seed growing algorithm previously described [DOI: 10.1097/JTO.0b013e3182843721]. The database was randomly split into an 80%-20% development and hold-out set prior to ML modeling. Feature selection for variables was conducted utilizing Elastic Net Regression, which provides a balanced approach between Lasso and Ridge regression. A logistic regression model was developed using the selected features and ground truth information of the training cohort. The true performance of the model was tested on the hold-out set. Results 160 radiomics and 22 clinical features were extracted. Elastic Net regression identified 44 features that contributed the most. Some of these clinical features include presence of satellite lesions, smoking history, and presence of positive IgG coccidiomycosis serology. The AUC of the ROC curve for the hold-out test set was 0.86. Conclusions The clinical features of significance that emerged during multimodal modeling are explainable. Designing explainable ML models is important, especially in the Southwestern US where Coccidioides nodules can appear to be malignant to both human experts and ML models. Although coccidioidomycosis serologies alone lack sensitivity and specificity to diagnose Coccidioides nodules, integrating them with a multimodal ML model can improve differentiation between benign Coccidioides and malignant adenocarcinoma nodules. This abstract is funded by: None

Abstract Alveolar macrophages (AMs) are critical for host defense and immunity against Pneumocystis. Through surface-associated receptors, AMs recognize, phagocytose, and kill Pneumocystis organisms, producing cytokines that promote pathogen clearance but also contribute to lung inflammation and injury. The EphA2 receptor, a member of the receptor tyrosine kinase (RTK) family, contains an extracellular region that binds activating ligands (Ephrins) and, more recently, has been shown to bind fungal β-glucans, leading to activation of its intracellular tyrosine kinase domain. The EphA2 signaling pathway has important roles in cancer, angiogenesis, and inflammatory diseases such as atherosclerosis. EphA2 is highly expressed in epithelial and endothelial cells, and emerging evidence indicates that it serves as a critical regulator of host defense against bacterial, viral, and fungal pathogens. In oropharyngeal candidiasis, EphA2 was identified as a receptor for fungal β-glucans in oral epithelial cells, mediating proinflammatory signaling through signal transducer and activator of transcription 3 (STAT3) and mitogen-activated protein kinase (MAPK) pathways. Our laboratory recently demonstrated that EphA2 on lung epithelial cells binds Pneumocystis organisms, and that β-glucans from these organisms activate downstream signaling leading to IL-6 release. However, the role of EphA2 signaling in AM immune responses during Pneumocystis pneumonia (PCP) remains unknown. To address this, we examined EphA2-deficient (EphA2−/−) AMs stimulated with Pneumocystis β-glucans. Compared to wild-type AMs, EphA2−/− AMs exhibited significantly decreased production of proinflammatory cytokines, including IL-6 and TNF-α. In CD4-depleted mouse models of PCP, EphA2−/− lungs displayed markedly reduced levels of inflammatory cytokines and increased organism burdens, indicating impaired immune activation but preserved fungal persistence. Importantly, in CD4-depleted mice undergoing standard trimethoprim-sulfamethoxazole (TMP/SMX) therapy, concurrent administration of the EphA2 small-molecule inhibitor ALW-II-41-27 significantly reduced lung TNF-α responses without evidence of toxicity. These findings demonstrate that EphA2 receptor signaling plays a critical role in regulating macrophage-mediated inflammatory responses to Pneumocystis. While EphA2 deficiency dampens pulmonary inflammation, it also compromises fungal clearance, highlighting a delicate balance between host defense and immunopathology. Our data suggest that targeted modulation of EphA2 signaling during anti-Pneumocystis therapy may provide a novel adjunctive approach to reduce detrimental lung inflammation associated with PCP, potentially improving outcomes in affected patients. This abstract is funded by: NIH R01 HL-62150