Collaborative Database Research Articles

U-shaped correlation of lymphocyte count with all-cause hospital mortality in sepsis and septic shock patients: a MIMIC-IV and eICU-CRD database study

BackgroundIn sepsis, the relationship between lymphocyte counts and patient outcomes is complex. Lymphocytopenia and lymphocytosis significantly influence survival, illustrating the dual functionality of lymphocytes in responding to infections. This study investigates this complex interaction, focusing on how variations in lymphocyte counts correlate with all-cause hospital mortality among sepsis patients.MethodsThis retrospective cohort study analyzed data from two extensive critical care databases: the Medical Information Mart for Intensive Care IV 2.0 (MIMIC-IV 2.0) from Beth Israel Deaconess Medical Center, Boston, Massachusetts, and the eICU Collaborative Research Database (eICU-CRD), which was Multi-center database from over 200 hospitals across the United States conducted by Philips eICU Research Institute. We included adult patients aged 18 years and older who met the Sepsis-3 criteria, characterized by documented or suspected infection and a Sequential Organ Failure Assessment (SOFA) score of 2 or higher. Sepsis patients were categorized into quartiles based on lymphocyte counts. The primary outcome was all-cause mortality in the hospital, with 90 and 60-day all-cause mortality as the secondary outcomes. Univariable and multivariable Cox proportional hazard regressions were utilized to assess lymphocyte counts' impact on hospital mortality. An adjusted restricted cubic spline (RCS) analysis was performed to elucidate this relationship further. Subgroup analyses were also conducted to explore the association across various comorbidity groups among sepsis and septic shock patients.ResultsOur study included 37,054 patients, with an observed in-hospital mortality rate of 16.6%. Univariable and multivariable Cox proportional hazard regression models showed that lymphocyte counts were independently associated with in-hospital mortality (HR = 1.04, P < 0.01; HR = 1.06, P < 0.01). RCS regression analysis revealed a U-shaped relationship between lymphocyte levels and hospital mortality risk in sepsis and septic shock patients (P for overall < 0.001, P for nonliner < 0.01; P for overall = 0.002, P for nonliner = 0.014). Subgroup analyses revealed that elevated lymphocyte counts correlated with increased hospital mortality among sepsis patients with liver disease and requiring renal replacement therapy (P for overall = 0.021, P for nonliner = 0.158; P for overall = 0.025, P for nonliner = 0.759). These findings suggest that lymphocytes may have enhanced prognostic value in specific subsets of critically ill sepsis patients.ConclusionOur findings demonstrate that lymphocyte counts are a significant independent predictor of hospital mortality in sepsis and septic shock patients. We observed a U-shaped association between lymphocyte levels and mortality risk, indicating that high and low counts are linked to increased mortality. This result highlights the complex role of lymphocytes in sepsis outcomes and suggests the need for further investigation into the underlying mechanisms and potential therapeutic approaches. Integrating lymphocyte count assessment into risk stratification algorithms and clinical decision support tools could enhance the early identification of high-risk sepsis patients.

Prognostic implications of margin status in association with systemic treatment in a cohort study of patients with resection of colorectal liver metastases.

This study investigates the impact of margin status after colorectal liver metastasis (CLM) resection on outcomes of patients after neoadjuvant treatment versus those who underwent upfront resection. An international collaborative database of CLM patients who underwent surgical resection was used. Proportional hazard regression models were created for single and multivariable models to assess the relationship between independent measures and median overall survival (mOS). R1 was associated with worse OS in the neoadjuvant group (mOS: 51.8 m for R0 vs. 26.0 m for R1; HR: 2.18). In the patients who underwent upfront surgery, R1 was not associated with OS. (mOS: 46.7 m for R0 vs. 42.6 m for R1). When patients with R1 in each group were stratified by adjuvant treatment, there was no significant difference in the neoadjuvant group, while in the upfront surgery group with R1, adjuvant treatment was associated with significant improvement in OS (mOS: 42.6 m for adjuvant vs. 25.0 m for no adjuvant treatment; HR: 0.21). R1 is associated with worse outcomes in the patients who receive neoadjuvant treatment with no significant improvement with the addition of adjuvant therapy, likely representing an aggressive tumor biology. R1 did not impact OS in patients with upfront surgery who received postoperative chemotherapy.

Racial Disparities in Pulse Oximetry, in COVID-19 and ICU Settings.

This study aimed to assess the impact of race on pulse oximetry reliability, taking into account Spo2 ranges, COVID-19 diagnosis, and ICU admission. Retrospective cohort study covering admissions from January 2020 to April 2024. National COVID Cohort Collaborative (N3C) database, consisting of electronic health records from 80 U.S. institutions. Patients were selected from the N3C database based on the availability of data on self-identified race and both pulse oximetry estimated Spo2 and Sao2. Subgroups included patients in ICU and non-ICU settings, with or without a diagnosis of COVID-19 disease. None. The agreement between Spo2 and Sao2 was assessed across racial groups (American Indian or Alaska Native, Asian, Black, Hispanic or Latino, Pacific Islander, and White). Each patient's initial Sao2 measurement was matched with the closest Spo2 values recorded within the preceding 10-minute time frame. The risk of hidden hypoxemia (Spo2 ≥ 88% but Sao2 < 88%) was determined for various Spo2 ranges, races, and clinical scenarios. We used a generalized logistic mixed-effects model to evaluate the impact of relevant variables, such as COVID-19, ICU admission, age, sex, race, and Spo2, on the risk of hidden hypoxemia, while accounting for the random effects within each hospital. A total of 80,541 patients were included, consisting of 596 American Indian or Alaska Native, 2,729 Asian, 11,889 Black, 13,154 Hispanic or Latino, 221 Pacific Islander, and 51,952 White individuals. Discrepancies between Spo2 and Sao2 were observed across all racial groups, with the most pronounced bias in Black patients. Hidden hypoxemia rates were higher in Black patients across all Spo2 subgroups, for all clinical scenarios. The odds of hidden hypoxemia were higher for Black and Hispanic or Latino patients and for those with COVID-19 disease. Race significantly impacts pulse oximetry reliability. Not only Black and Hispanic or Latino patients were at higher risk for hidden hypoxemia, but also those admitted with a COVID-19 diagnosis. Future in-depth explorations into the underlying causes and potential solutions are needed.

Association of changes in HerQLes scores with objective hernia outcomes: an analysis of the ACHQC database.

There are both objective and subjective measures of success following ventral hernia repair (VHR). Using the Abdominal Core Health Quality Collaborative (ACHQC) database, we sought to determine if there is an association between 30-day wound events (objective) and changes in the hernia-related quality-of-life (HerQLes) scores, (subjective). We hypothesized that patients who do not experience a 30-day wound event have a greater improvement in their HerQLes score over the short-term. All adult patients who underwent VHR with 30-day follow-up data available between 2013 and 2022 were identified within the ACHQC database. The 30-day wound events included surgical site infection (SSI), surgical site occurrence (SSO), and SSO requiring procedural intervention (SSOPI). The association between 30-day wound events and changes in HerQLes scores was measured using propensity matched score analysis. Further, regression analysis was used to determine if an improvement in HerQLes score at 30-days postoperatively was associated with the likelihood of experiencing a 30-day wound event. Following a 3:1 matched analysis, 17,796 patients were available for analysis; 4449 (25%) patients experienced a 30-day wound event. The most common SSI was a superficial SSI and the most common SSO was a seroma. A 10-point improvement in the HerQLes score was statistically associated with a 3% decrease in SSI and a 4% decrease in the odds of experiencing an SSO. While not statistically significant, a 10-point improvement in the HerQLes score was associated with a 2.4% decrease in the odds of experiencing an SSOPI. Subjective and objective measures of success following VHR seem to be correlated with one another over the short-term. Additional studies are needed to determine if this correlation exists with other subjective and objective measures of success and to determine if these correlations persist over the long-term. If present, these associations may help to guide patient counseling as experiencing a postoperative wound event following ventral hernia repair may not be detrimental to their quality-of-life over the long-term.

Defect width and operative time are the most significant predictors of venous thromboembolism in ventral hernia repair

BackgroundVenous thromboembolism is a significant cause of morbidity and mortality among patients undergoing ventral hernia repair. Several risk-assessment models have been developed to predict venous thromboembolism risk for various surgical procedures. However, these models do not include hernia-specific risk factors. Therefore, we sought to evaluate the predictors of venous thromboembolism in patients with a ventral hernia repair in a national hernia-specific database. MethodsThe Abdominal Core Health Quality Collaborative database was retrospectively queried for ventral hernia repair data. The ventral hernia repair procedures were divided into 2 groups on the basis of whether the patients developed perioperative venous thromboembolism. Baseline and operative characteristics and perioperative outcomes were compared between the 2 groups. The logistic regression was used to assess the association between different risk factors and venous thromboembolism. ResultsA total of 43,558 patients were included in the study. The 30-day venous thromboembolism rate among these patients was 0.62% (n = 269). Venous thromboembolism was associated with greater 30-day readmission (odds ratio, 4.29; 95% confidence interval, 3.20–5.76), reoperation (odds ratio, 3.97, 95% confidence interval, 2.63–6.00), and mortality rates (odds ratio, 5.95, 95% confidence interval, 2.62–13.48), all P < .001. Bivariate analysis identified a statistically significant association between several patient characteristics, operation, and hernia- related details; however, multivariate analysis revealed that that only hernia width (each increasing centimeter: odds ratio, 1.07, 95% CI, 1.05–1.09) and operative time compared with 0–59 minutes (60–119 minutes: odds ratio, 4.23; 95% confidence interval, 1.86–9.62; 120–179 minutes: odds ratio, 5.78; 95% confidence interval, 2.57–13.0; 180–239 minutes: odds ratio, 8.01; 95% confidence interval, 3.54–18.10; and 240+ minutes: odds ratio, 17.4; 95% confidence interval, 7.88–38.37) were significantly associated with venous thromboembolism risk. ConclusionsVenous thromboembolism is an uncommon complication of ventral hernia repair. It is associated with increased readmission, reoperation, and mortality rates. Larger defect width and operative time are most predictive of perioperative venous thromboembolism risk for patients undergoing ventral hernia repair.

Interpretable machine learning model for early prediction of acute kidney injury in patients with rhabdomyolysis

Abstract Background Rhabdomyolysis (RM) is a complex set of clinical syndromes. RM-induced acute kidney injury (AKI) is a common illness in war and military operations. This study aimed to develop an interpretable and generalizable model for early AKI prediction in patients with RM. Methods Retrospective analyses were performed on 2 electronic medical record databases: the eICU Collaborative Research Database and the Medical Information Mart for Intensive Care III database. Data were extracted from the first 24 hours after patient admission. Data from the two datasets were merged for further analysis. The extreme gradient boosting (XGBoost) model with the Shapley additive explanation method (SHAP) was used to conduct early and interpretable predictions of AKI. Results The analysis included 938 eligible patients with RM. The XGBoost model exhibited superior performance (area under the receiver operating characteristic curve [AUC] = 0.767) compared to the other models (logistic regression, AUC = 0.711; support vector machine, AUC = 0.693; random forest, AUC = 0.728; and naive Bayesian, AUC = 0.700). Conclusion Although the XGBoost model performance could be improved from an absolute perspective, it provides better predictive performance than other models for estimating the AKI in patients with RM based on patient characteristics in the first 24 hours after admission to an intensive care unit. Furthermore, including SHAP to elucidate AKI-related factors enables individualized patient treatment, potentially leading to improved prognoses for patients with RM.

APRICOT-Mamba: Acuity Prediction in Intensive Care Unit (ICU): Development and Validation of a Stability, Transitions, and Life-Sustaining Therapies Prediction Model.

On average, more than 5 million patients are admitted to intensive care units (ICUs) in the US, with mortality rates ranging from 10 to 29%. The acuity state of patients in the ICU can quickly change from stable to unstable, sometimes leading to life-threatening conditions. Early detection of deteriorating conditions can assist in more timely interventions and improved survival rates. While Artificial Intelligence (AI)-based models show potential for assessing acuity in a more granular and automated manner, they typically use mortality as a proxy of acuity in the ICU. Furthermore, these methods do not determine the acuity state of a patient (i.e., stable or unstable), the transition between acuity states, or the need for life-sustaining therapies. In this study, we propose APRICOT-M (Acuity Prediction in Intensive Care Unit-Mamba), a 1M-parameter state space-based neural network to predict acuity state, transitions, and the need for life-sustaining therapies in real-time among ICU patients. The model integrates ICU data in the preceding four hours (including vital signs, laboratory results, assessment scores, and medications) and patient characteristics (age, sex, race, and comorbidities) to predict the acuity outcomes in the next four hours. Our state space-based model can process sparse and irregularly sampled data without manual imputation, thus reducing the noise in input data and increasing inference speed. The model was trained on data from 107,473 patients (142,062 ICU admissions) from 55 hospitals between 2014-2017 and validated externally on data from 74,901 patients (101,356 ICU admissions) from 143 hospitals. Additionally, it was validated temporally on data from 12,927 patients (15,940 ICU admissions) from one hospital in 2018-2019 and prospectively on data from 215 patients (369 ICU admissions) from one hospital in 2021-2023. Three datasets were used for training and evaluation: the University of Florida Health (UFH) dataset, the electronic ICU Collaborative Research Database (eICU), and the Medical Information Mart for Intensive Care (MIMIC)-IV dataset. APRICOT-M significantly outperforms the baseline acuity assessment, Sequential Organ Failure Assessment (SOFA), for mortality prediction in both external (AUROC 0.95 CI: 0.94-0.95 compared to 0.78 CI: 0.78-0.79) and prospective (AUROC 0.99 CI: 0.97-1.00 compared to 0.80 CI: 0.65-0.92) cohorts, as well as for instability prediction (external AUROC 0.75 CI: 0.74-0.75 compared to 0.51 CI: 0.51-0.51, and prospective AUROC 0.69 CI: 0.64-0.74 compared to 0.53 CI: 0.50-0.57). This tool has the potential to help clinicians make timely interventions by predicting the transition between acuity states and decision-making on life-sustaining within the next four hours in the ICU.

Risk factors, incidence, and outcomes of neonatal respiratory extracorporeal membrane oxygenation including association with therapeutic hypothermia in California during 2013-2020.

To evaluate the incidence, indications and outcomes with neonatal extracorporeal membrane oxygenation (ECMO) and its association with therapeutic hypothermia (TH) among infants undergoing invasive mechanical ventilation (IMV) in California during 2013-2020. We analyzed data on neonates ≥34 weeks gestation with ≥4 h of IMV over an 8-year period (2013-2020) from the California Perinatal Quality Care Collaborative (CPQCC) database. Between 2013 and 2020, the ranges for utilization of iNO (13.9 to 17.2%), ECMO (2.1 to 2.5%), TH (10.2 to 15.7%) and TH + ECMO (0.4 to 0.8%) were observed. The most common association with neonatal ECMO was TH (148 cases, OR 3.2, 95% CI 2.6-4.3, p < 0.01). The combination of meconium aspiration syndrome (MAS) and hypoxic ischemic encephalopathy (HIE) increased risk of iNO and ECMO use (OR 11.3, 1.5-86.9), p = 0.02). Ventilated infants ≥34 weeks gestational age undergoing TH are at risk for iNO/ECMO use and need close monitoring.

An early sepsis prediction model utilizing machine learning and unbalanced data processing in a clinical context

BackgroundEarly and accurate diagnoses of sepsis patients are essential to reduce the mortality. However, the sepsis is still diagnosed in a traditional way in China despite the increasing number of related studies, which may to some extent lead to delays in the treatment. MethodsThe study included 2,385 patients, including 364 with sepsis, collected from the First Affiliated Hospital of Anhui Medical University and partner hospitals from April to July 2022. External validation was conducted using the MIMIC-III database (over 60,000 patients from 2001 to 2012) and the eICU Collaborative Research Database (139,000 patients from 2014 to 2015). Multiple algorithm models, along with the SHapley Additive exPlanations (SHAP) analysis, are applied to explore the main risk factors for the accurate prediction of the sepsis. Multiple Imputations for filling missing data and the Synthetic Minority Oversampling (SMOTE) balancing method for balancing data are used for the data processing. ResultEighteen diagnostic features are used in the predictive model for early sepsis. The Random Forest model has the best performance among all the models, with an Area Under the Curve (AUC) of 87% and an F1-score (F1) of 77%. Moreover, the interpretation from the SHAP analysis is generally consistent with the current clinical situation. ConclusionThe study revealed the relationship between these 18 clinical features and diagnostic outcomes. The results indicate that patients with laboratory values of Systolic Blood Pressure, Albumin, and Heart Rate exceeding certain thresholds are at a high likelihood of developing sepsis.

The neutrophil-to-lymphocyte ratio levels over time correlate to all-cause hospital mortality in sepsis

ObjectiveThis research aims to investigate the prognosis value using the time-weighted average neutrophil-to-lymphocyte ratio (TWA-NLR) for predicting all-cause hospital mortality among sepsis patients. Data were analyzed through the use of the eICU Collaborative Research Database (eICU-CRD 2.0) as well as Medical Information Mart for Intensive Care IV 2.2 (MIMIC-IV 2.2). MethodsSeptic patients from both eICU-CRD 2.0 as well as MIMIC-IV 2.2 databases were included. The neutrophil-to-lymphocyte ratios (NLR) were available for analysis, utilizing complete blood counts obtained on days one, four, and seven following ICU admission. The TWA-NLR was computed at the end of the seven days, and patients were then stratified based on TWA-NLR thresholds. 90-day all-cause mortality during hospitalization was the primary objective, with 60-day all-cause hospital mortality as a secondary objective. The correlation between TWA-NLR and sepsis patients' primary outcome was analyzed using univariable and multivariable Cox proportional hazard regressions. A restricted cubic spline (RCS) analysis was conducted in an attempt to confirm this association further, and subgroup analyses were employed to evaluate the correlation across various comorbidity groups. Results3921 patients were included from the eICU-CRD 2.0, and the hospital mortality rate was 20.8 %. Both multivariable as well as univariable Cox proportional hazard regression analyses revealed that TWA-NLR was independently correlated with 90-day all-cause hospital mortality, yielding a hazard ratio (HR) of 1.02 (95 % CI 1.01–1.02, P-value<0.01) as well as 1.12 (95 % CI 1.01–1.15, P-value<0.01), respectively. The RCS analysis demonstrated a significant nonlinear relationship between TWA-NLR and 90-day all-cause hospital mortality risk. The study subjects were divided into higher (>10.5) and lower (≤10.5) TWA-NLR cohorts. A significantly decreased incidence of 90-day all-cause hospital mortality (HR = 0.56, 95 % CI 0.48–0.64, P-value<0.01) and longer median survival time (40 days vs 24 days, P-value<0.05) were observed in the lower TWA-NLR cohort. However, septic patients with chronic pulmonary (interaction of P-value = 0.009) or renal disease (interaction of P-value = 0.008) exhibited significant interactive associations between TWA-NLR and 90-day all-cause hospital mortality, suggesting the predictive power of TWA-NLR may be limited in these subgroups. The MIMIC-IV 2.2 was utilized as a validation cohort and exhibited a similar pattern. ConclusionOur findings suggest that TWA-NLR is a powerful and independent prognostic indicator for 90-day all-cause hospital mortality among septic patients, and the TWA-NLR cutoff value may prove a useful method for identifying high-risk septic patients.

The Comparison of Newly Diagnosed Invasive Breast Patient Cohorts in Genomics Evidence Neoplasia Information Exchange Biopharma Collaborative (GENIE-BPC) and Other Real-World Databases.

Oncology databases that integrate genomic and clinical data have become valuable resources for precision medicine. However, the generalizability of these databases has not been comprehensively assessed. To describe the demographics, clinical characteristics, treatments, and overall survival of breast cancer cohorts in GENIE-BPC and three other databases. This study utilized GENIE-BPC, SEER, SEER-Medicare, and Merative MarketScan Research Databases. Women with invasive breast cancer were identified through EHR, cancer registries or ICD-9/10-CM codes. The ages were 18+ years or per database requirement. Treatments were based on EHR or HCPCS/NDC codes in claims. Overall survival was estimated as time from diagnosis to death. Of female breast cancer patients in GENIE-BPC (n = 775), SEER (n = 548 336), SEER-Medicare (n = 68 914), and Marketscan (n = 109 499) databases, the median ages at initial diagnosis were 44, 62, 74, and 57 years, respectively. A greater proportion of patients in GENIE-BPC, compared to SEER/SEER-Medicare, had higher nuclear grades (%III-%IV: 57% vs. 26%/24%), advanced disease stage (%IV: 25.3% vs. 5%/3.6%), percent of triple negative breast cancer (19.7% vs. 10.2%/8.5%), and receipt of chemotherapy (85.0% vs. NA/22.3%). The 1-, 3-, and 5-year overall survival rates were lower in GENIE-BPC (78.5%, 60.5%, 55.5%) than in SEER (95.8%, 89.5%, 85.5%) and SEER-Medicare (91.6%, 81.4%, 75.0%). Breast cancer patients in GENIE-BPC were younger, had more advanced disease, had a higher proportion of triple negative breast cancer and recipients of chemotherapy, and had poorer overall survival. Researchers must use statistical adjustment when extrapolating results (e.g., biomarker prevalence) from GENIE-BPC to the larger breast cancer population.

An Unsupervised Error Detection Methodology for Detecting Mislabels in Healthcare Analytics.

Medical datasets may be imbalanced and contain errors due to subjective test results and clinical variability. The poor quality of original data affects classification accuracy and reliability. Hence, detecting abnormal samples in the dataset can help clinicians make better decisions. In this study, we propose an unsupervised error detection method using patterns discovered by the Pattern Discovery and Disentanglement (PDD) model, developed in our earlier work. Applied to the large data, the eICU Collaborative Research Database for sepsis risk assessment, the proposed algorithm can effectively discover statistically significant association patterns, generate an interpretable knowledge base for interpretability, cluster samples in an unsupervised learning manner, and detect abnormal samples from the dataset. As shown in the experimental result, our method outperformed K-Means by 38% on the full dataset and 47% on the reduced dataset for unsupervised clustering. Multiple supervised classifiers improve accuracy by an average of 4% after removing abnormal samples by the proposed error detection approach. Therefore, the proposed algorithm provides a robust and practical solution for unsupervised clustering and error detection in healthcare data.

Predicting Tracheostomy Need on Admission to the Intensive Care Unit-A Multicenter Machine Learning Analysis.

It is difficult to predict which mechanically ventilated patients will ultimately require a tracheostomy which further predisposes them to unnecessary spontaneous breathing trials, additional time on the ventilator, increased costs, and further ventilation-related complications such as subglottic stenosis. In this study, we aimed to develop a machine learning tool to predict which patients need a tracheostomy at the onset of admission to the intensive care unit (ICU). Retrospective Cohort Study. Multicenter Study of 335 Intensive Care Units between 2014 and 2015. The eICU Collaborative Research Database (eICU-CRD) was utilized to obtain the patient cohort. Inclusion criteria included: (1) Age >18 years and (2) ICU admission requiring mechanical ventilation. The primary outcome of interest included tracheostomy assessed via a binary classification model. Models included logistic regression (LR), random forest (RF), and Extreme Gradient Boosting (XGBoost). Of 38,508 invasively mechanically ventilated patients, 1605 patients underwent a tracheostomy. The XGBoost, RF, and LR models had fair performances at an AUROC 0.794, 0.780, and 0.775 respectively. Limiting the XGBoost model to 20 features out of 331, a minimal reduction in performance was observed with an AUROC of 0.778. Using Shapley Additive Explanations, the top features were an admission diagnosis of pneumonia or sepsis and comorbidity of chronic respiratory failure. Our machine learning model accurately predicts the probability that a patient will eventually require a tracheostomy upon ICU admission, and upon prospective validation, we have the potential to institute earlier interventions and reduce the complications of prolonged ventilation.

Blood pressure response index and clinical outcomes in patients with septic shock: a multicenter cohort study

Sepsis is a leading cause of mortality in intensive care units and vasoactive drugs are widely used in septic patients. The cardiovascular response of septic shock patients during resuscitation therapies and the relationship of the cardiovascular response and clinical outcome has not been clearly described. We included adult patients admitted to the ICU with sepsis from Peking Union Medical College Hospital (internal), Medical Information Mart for Intensive Care IV (MIMIC-IV) and eICU Collaborative Research Database (eICU-CRD). The Blood Pressure Response Index (BPRI) was defined as the ratio between the mean arterial pressure and the vasoactive-inotropic score. BRRI was compared with existing risk scores on predicting in-hospital death. The relationship between BPRI and in-hospital mortality was calculated. A XGBoost's machine learning model identified the features that influence short-term changes in BPRI. There were 2139, 9455, and 4202 patients in the internal, MIMIC-IV and eICU-CRD cohorts, respectively. BPRI had a better AUROC for predicting in-hospital mortality than SOFA (0.78 vs. 0.73, p=0.01) and APS (0.78 vs. 0.74, p=0.03) in the internal cohort. The estimated odds ratio for death per unit decrease in BPRI was 1.32 (95% CI 1.20-1.45) when BPRI was below 7.1 vs. 0.99 (95% CI 0.97-1.01) when BPRI was above 7.1 in the internal cohort; similar relationships were found in MIMIC-IV and eICU-CRD. Respiratory support and latest cumulative 12-h fluid balance were intervention-related features influencing BPRI. BPRI is an easy, rapid, precise indicator of the response of patients with septic shock to vasoactive drugs. It is a comparable and even better predictor of prognosis than SOFA and APS in sepsis and it is simpler and more convenient in use. The application of BPRI could help clinicians identify potentially at-risk patients and provide clues for treatment. Fundings for the Beijing Municipal Natural Science Foundation; the National High Level Hospital Clinical Research Funding; the CAMS Innovation Fund for Medical Sciences (CIFMS) from Chinese Academy of Medical Sciences and the National Key R&D Program of China, Ministry of Science and Technology of the People's Republic of China.

Fine-grained subphenotypes in acute kidney injury populations based on deep clustering: Derivation and interpretation

BackgroundAcute kidney injury (AKI) is associated with increased mortality in critically ill patients. Due to differences in the etiology and pathophysiological mechanism, the current AKI criteria put it an embarrassment to evaluate clinical therapy and prognosis. ObjectiveWe aimed to identify subphenotypes based on routinely collected clinical data to expose the unique pathophysiologic patterns. MethodsA retrospective study was conducted based on the Medical Information Mart for Intensive Care IV (MIMIC-IV) and the eICU Collaborative Research Database (eICU-CRD), and a deep clustering approach was conducted to derive subphenotypes. We conducted further analysis to uncover the underlying clinical patterns and interpret the subphenotype derivation. ResultsWe studied 14,189 and 19,382 patients with AKI within 48 h of ICU admission in the two datasets, respectively. Through our approach, we identified seven distinct AKI subphenotypes with mortality heterogeneity in each cohort. These subphenotypes displayed significant variations in demographics, comorbidities, levels of laboratory measurements, and survival patterns. Notably, the subphenotypes could not be effectively characterized using the Kidney Disease: Improving Global Outcomes (KDIGO) criteria alone. Therefore, we uncovered the unique underlying characteristics of each subphenotype through model-based interpretation. To assess the usability of the subphenotypes, we conducted an evaluation, which yielded a micro-Area Under the Receiver Operating Characteristic (AUROC) of 0.81 in the single-center cohort and 0.83 in the multi-center cohort within 48-hour of admission. ConclusionWe derived highly characteristic, interpretable, and usable AKI subphenotypes that exhibited superior prognostic values.

Six Square Miles of Urban America: Association Between Firearm Discharge, Injury, and Fatality.

Despite the increase in firearm injury observed across the country, significant gaps remain relevant to our understanding of how firearm exposure translates to injury. Using acoustic gunshot detection and a collaborative hospital and law enforcement firearm injury database, we sought to identify the relationship between firearm discharge and injury over time. From 2018 to 2021, instances of firearm discharge captured via acoustic detection in 6 square miles of Louisville, KY, were merged with data from the collaborative firearm injury database. Key outcomes included the total number of rounds fired, injury and fatality rates per round, and the percentage of rounds discharged from automatic weapons and high-capacity magazines. During the study period, 54,397 rounds of ammunition were discharged resulting in 914 injuries, 435 hospital admissions, 2,442 hospital days, 155 emergent operations, and 180 fatalities. For each round of ammunition fired, the risk of injury and fatality was 1.7% and 0.3%, respectively. The total number of rounds fired per month nearly tripled (614 vs 1,623, p < 0.001) leading to increased injury (15 vs 37, p < 0.001) and fatality (3 vs 7, p < 0.001). The percentage of rounds fired from automatic weapons (0 vs 6.8%, p < 0.001) and high-capacity magazines (7.6 vs 28.9%, p < 0.001) increased over time. The increased burden of firearm injury is related to an overall increase in firearm exposure as measured by the total number of rounds discharged. High-capacity magazines and automatic weaponry are being used with increasing frequency in urban America.

Postoperative Cesarean Section Pain Management Using Transversus Abdominis Plane Block Versus Intrathecal Morphine: A Systematic Review and Meta-analysis

PurposeCompare and evaluate the effectiveness of transversus abdominis plane (TAP) block versus intrathecal morphine (ITM) on elective postcesarean section pain, opioid consumption, and related side effects. DesignSystematic review and meta-analysis. MethodsA search for evidence was conducted in PubMed, Google Scholar, CINAHL, Cochrane Collaboration Database, UpToDate, Health Source, and gray literature. Only randomized controlled trials (RCTs) were included in the study. The methodological quality of evidence assessment was conducted using the Risk of Bias and Grades of Recommendation, Assessment, Development, and Evaluation system. The meta-analysis used Review Manager (RevMan 5.4, The Cochrane Collaboration). FindingsA total of 11 RCTs involving 1,129 patients were analyzed. Compared to ITM, TAP has a similar effect on static (mean difference [MD]; 0.37; 95% confidence interval [CI], −0.04 to 0.79; P = .08) and dynamic pain scores (MD, 0.43; 95% CI, −0.06 to 0.92; P = .09) within the first 48 hours after surgery. Additionally, the TAP block had a lower incidence of postoperative nausea and vomiting (risk ratio, 0.45; 95% CI, 0.31 to 0.66; P < .0001) and increased opioid consumption (MD, 6.78; 95% CI, 3.79 to 9.77; P < .00001). Overall, TAP block and ITM did not differ in the time to first to rescue analgesia, incidence of sedation, and pruritus. ConclusionsEvidence suggests that TAP blocks are equivalent to ITM in pain scores and more effective at lowering the incidence of postoperative nausea and vomiting, yet ITM has been shown to be more effective in reducing postoperative opioid consumption.

Characterizing phenotypes and clinical and health utilization associations of young people with chronic pain: latent class analysis using the electronic Persistent Pain Outcomes Collaboration database.

Using the Australiasian electronic Persistent Pain Outcomes Collaboration, a binational pain registry collecting standardized clinical data from paediatric ePPOC (PaedsePPOC) and adult pain services (AdultePPOC), we explored and characterized nationally representative chronic pain phenotypes and associations with clinical and sociodemographic factors, health care utilization, and medicine use of young people. Young people ≥15.0 and <25.0 years captured in PaedePPOC and AdultePPOC Australian data registry were included. Data from 68 adult and 12 paediatric pain services for a 5-year period January 2018 to December 2022 (first episode, including treatment information) were analysed. Unsupervised latent class analysis was applied to explore the existence of distinct pain phenotypes, with separate models for both services. A 3-phenotype model was selected from both paediatric and adult ePPOC data, with 693 and 3518 young people included, respectively (at least one valid indicator variable). Indicator variables for paediatric models were as follows: pain severity, functional disability (quasisurrogate "pain interference"), pain count, pain duration, pain-related worry (quasisurrogate "catastrophizing"), and emotional functioning; and, for adult models: pain severity, pain interference, pain catastrophizing, emotional functioning, and pain self-efficacy. From both services, 3 similar phenotypes emerged ("low," "moderate," "high"), characterized by an increasing symptom-severity gradient in multidimensional pain-related variables, showing meaningful differences across clinical and sociodemographic factors, health service utilization, and medicines use. Derived phenotypes point to the need for novel care models that differentially respond to the needs of distinct groups of young people, providing timely, targeted, age-appropriate care. To effectively scale such care, digital technologies can be leveraged to augment phenotype-informed clinical care.

The search for optimal nocturnal diurnal heart rate Index targets in ICU patients: a retrospective observational study from large ICU database.

Circadian rhythms play a crucial role in cardiovascular health, with the nocturnal diurnal heart rate index (NDHRI) reflecting significant circadian variations. However, the optimal NDHRI target in Intensive Care Unit (ICU) patients remains undefined. This study aims to establish an evidence-based NDHRI target range and assess its association with mortality. Data from the eICU Collaborative Research Database (n = 32,412) were analyzed. NDHRI was calculated by dividing cumulative nighttime heart rate area by daytime area. Generalized additive models (GAMs) explored the non-linear relationship between mean NDHRI and mortality, adjusting for confounders. Subgroup analyses were conducted based on ethnicity, ICU type, and comorbidities. A U-shaped association was observed between hospital mortality and mean NDHRI (P < 0.001). The optimal NDHRI range (40.0%-45.0%) demonstrated the lowest mortality rates. The duration spent within this range correlated inversely with mortality (P < 0.001). Subgroup analyses consistently supported these findings across diverse patient profiles. Our findings suggest an association between maintaining NDHRI within the 40.0%-45.0% range and lower mortality rates in critically ill patients, highlighting the potential utility of monitoring circadian heart rate variations in the ICU. Further research and future randomized controlled trials are essential to confirm causality and should consider this NDHRI range as a pivotal reference target.

Predicting the Arousal and Valence Values of Emotional States Using Learned, Predesigned, and Deep Visual Features.

The cognitive state of a person can be categorized using the circumplex model of emotional states, a continuous model of two dimensions: arousal and valence. The purpose of this research is to select a machine learning model(s) to be integrated into a virtual reality (VR) system that runs cognitive remediation exercises for people with mental health disorders. As such, the prediction of emotional states is essential to customize treatments for those individuals. We exploit the Remote Collaborative and Affective Interactions (RECOLA) database to predict arousal and valence values using machine learning techniques. RECOLA includes audio, video, and physiological recordings of interactions between human participants. To allow learners to focus on the most relevant data, features are extracted from raw data. Such features can be predesigned, learned, or extracted implicitly using deep learners. Our previous work on video recordings focused on predesigned and learned visual features. In this paper, we extend our work onto deep visual features. Our deep visual features are extracted using the MobileNet-v2 convolutional neural network (CNN) that we previously trained on RECOLA's video frames of full/half faces. As the final purpose of our work is to integrate our solution into a practical VR application using head-mounted displays, we experimented with half faces as a proof of concept. The extracted deep features were then used to predict arousal and valence values via optimizable ensemble regression. We also fused the extracted visual features with the predesigned visual features and predicted arousal and valence values using the combined feature set. In an attempt to enhance our prediction performance, we further fused the predictions of the optimizable ensemble model with the predictions of the MobileNet-v2 model. After decision fusion, we achieved a root mean squared error (RMSE) of 0.1140, a Pearson's correlation coefficient (PCC) of 0.8000, and a concordance correlation coefficient (CCC) of 0.7868 on arousal predictions. We achieved an RMSE of 0.0790, a PCC of 0.7904, and a CCC of 0.7645 on valence predictions.