Mortality Risk Stratification Research Articles

Stratifying Mortality Risk in Intensive Care: A Comprehensive Analysis Using Cluster Analysis and Classification and Regression Tree Algorithms

BackgroundMachine learning (ML) can be promising for stratifying patients into homogeneous groups and assessing mortality based on score combination. Using ML, we compared mortality prediction performance for clustered and non-clustered models and tried to develop a simple decision algorithm to predict the patient’s cluster membership with classification and regression trees (CART).MethodsRetrospective study involving patients requiring ICU admission (1st January 2011–16th September 2022). Clusters were identified by combining Charlson Comorbidity Index (CCI) plus Simplified Acute Physiology Score II (SAPS II) or Sequential Organ Failure Assessment (SOFA). Intercluster and survival analyses were performed. We analyzed the relationship with mortality with multivariate logistic regressions and receiver operating characteristic curves (ROC) for models with and without clusters. Nested models were compared with Likelihood Ratio Tests (LRT). Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were compared for non-nested models. With the best model, we used CART to build a decision tree for patient’s membership.ResultsOur sample consisted of 2605 patients (mortality 59.7%). For both score combinations, we identified two clusters (A and B for CCI + SAPS II, α and β for CCI + SOFA). Belonging to cluster B/β was associated with shorter survival times (Peto-Peto p-values < 0.0001) and increased mortality (Odds-ratio 4.65 and 5.44, respectively). According to LRT and ROC analysis, clustered models performed better, and CCI + SOFA showed the lowest AIC and BIC values (AIC = 3021.21, BIC = 3132.65). Using CART (β cluster positive case) the accuracy of the decision tree was 94.8%.ConclusionClustered models significantly improved mortality prediction. The CCI + SOFA clustered model showed the best balance between complexity and data fit and should be preferred. Developing a user-friendly decision-making algorithm for cluster membership with CART showed high accuracy. Further validation studies are needed to confirm these findings.

Development of interpretable machine learning models to predict in-hospital prognosis of acute heart failure patients.

In recent years, there has been remarkable development in machine learning (ML) models, showing a trend towards high prediction performance. ML models with high prediction performance often become structurally complex and are frequently perceived as black boxes, hindering intuitive interpretation of the prediction results. We aimed to develop ML models with high prediction performance, interpretability, and superior risk stratification to predict in-hospital mortality and worsening heart failure (WHF) in patients with acute heart failure (AHF). Based on the Kyoto Congestive Heart Failure registry, which enrolled 4056 patients with AHF, we developed prediction models for in-hospital mortality and WHF using information obtained on the first day of admission (demographics, physical examination, blood test results, etc.). After excluding 16 patients who died on the first or second day of admission, the original dataset (n=4040) was split 4:1 into training (n=3232) and test datasets (n=808). Based on the training dataset, we developed three types of prediction models: (i) the classification and regression trees (CART) model; (ii) the random forest (RF) model; and (iii) the extreme gradient boosting (XGBoost) model. The performance of each model was evaluated using the test dataset, based on metrics including sensitivity, specificity, area under the receiver operating characteristic curve (AUC), Brier score, and calibration slope. For the complex structure of the XGBoost model, we performed SHapley Additive exPlanations (SHAP) analysis, classifying patients into interpretable clusters. In the original dataset, the proportion of females was 44.8% (1809/4040), and the average age was 77.9±12.0. The in-hospital mortality rate was 6.3% (255/4040) and the WHF rate was 22.3% (900/4040) in the total study population. In the in-hospital mortality prediction, the AUC for the XGBoost model was 0.816 [95% confidence interval (CI): 0.815-0.818], surpassing the AUC values for the CART model (0.683, 95% CI: 0.680-0.685) and the RF model (0.755, 95% CI: 0.753-0.757). Similarly, in the WHF prediction, the AUC for the XGBoost model was 0.766 (95% CI: 0.765-0.768), outperforming the AUC values for the CART model (0.688, 95% CI: 0.686-0.689) and the RF model (0.713, 95% CI: 0.711-0.714). In the XGBoost model, interpretable clusters were formed, and the rates of in-hospital mortality and WHF were similar among each cluster in both the training and test datasets. The XGBoost models with SHAP analysis provide high prediction performance, interpretability, and reproducible risk stratification for in-hospital mortality and WHF for patients with AHF.

Machine learning with electrocardiograms to optimize mortality risk stratification in patients with suspected acute coronary syndrome

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Heart, Lung, and Blood Institute (NHLBI) and National Center for Advancing Translational Sciences (NCATS). Background The importance of risk stratification in patients with suspected acute coronary syndrome (ACS) extends beyond diagnosis and immediate treatment. It influences the precision of care delivery and the allocation of resources to those at highest risk of adverse events. Purpose We sought to evaluate the prognostic value of electrocardiogram (ECG) feature-based machine learning models to risk-stratify long-term mortality in those with suspected ACS. Methods This was a multicenter prospective observational cohort study of consecutive, non-traumatic patients evaluated at the emergency department for suspected ACS. The derivation cohort included 4,015 patients from a University Medical Center (age 59±16 years, 47% women, 80% training with 10-fold cross-validation and 20% testing). Ascertainment of all-cause death was based on numerous data sources, including the Centers for Disease Control and Prevention’s National Death Index registry. We trained six machine learning methods for survival analysis using 73 morphological ECG features from presenting prehospital 10-second 12-lead ECGs. We used predicted risk scores in a variational Bayesian Gaussian mixture model to define three risk groups corresponding to low, moderate, and high-risk and compared the resulting classification performance against the HEART score. The external testing cohort included 3,095 patients from a University (age 59±15 years, 44% women). Results The mortality rate was 20.3% in the derivation cohort after an average follow-up period of 3.53 years (IQR 1.75 - 5.32). Extra Survival Trees outperformed other forecasting models during cross-validation. In the internal testing cohort, the derived risk groups were significantly predictive of survival (Kaplan-Meier log-rank test statistic = 121.14, p&amp;lt;0.001), outperforming the classification performance using the HEART score (Figure 1). Our machine learning-based risk stratification model detected &amp;gt;90% of death events missed by low-risk HEART score (Figure 2). Ruling-out the low-risk group, it achieved a negative predictive value of 93.4% with a sensitivity of 85.9% (compared to 89.0% and 75.0%, respectively, for the HEART score). This classification performance generalized well to our external testing cohort. Compared to those in the low-risk group, patients at moderate (OR = 3.62 [1.35-9.74]) and high (OR = 6.12 [2.38-15.75]) risk had significantly higher odds of 30-day cardiovascular death (n=109, 3.5%). Conclusions Using only features from the 10-second 12-lead ECG, we derived and externally validated a machine learning model that stratifies the mortality risk of patients during long-term follow up. This model outperformed current standard of care based on the HEART score. Future work should validate the potential of this decision-support tool in influencing treatment plans and resource allocation.Figure 1.Clustering performanceFigure 2.Sankey diagram

Development and validation of a novel sleep health score in the sleep heart health study

BackgroundThere is a lack of consensus in evaluating multidimensional sleep health, especially concerning its implication for mortality. A validated multidimensional sleep health score is the foundation of effective interventions. MethodsWe obtained data from 5706 participants in the Sleep Heart Health Study. First, random forest-recursive feature elimination algorithm was used to select potential predictive variables. Second, a sleep composite score was developed based on the regression coefficients from a Cox proportional hazards model evaluating the associations between selected sleep-related variables and mortality. Last, we validated the score by constructing Cox proportional hazards models to assess its association with mortality. ResultsThe mean age of participants was 63.2 years old, and 47.6% (2715/5706) were male. Six sleep variables, including average oxygen saturation (%), spindle density (C3), sleep efficiency (%), spindle density (C4), percentage of fast spindles (%) and percentage of rapid eye movement (%) were selected to construct this multidimensional sleep health score. The average sleep composite score in participants was 6.8 of 22 (lower is better). Participants with a one-point increase in sleep composite score had an 10% higher risk of death (hazard ratio = 1.10, 95% confidence interval: 1.08–1.12). ConclusionsThis study constructed and validated a novel multidimensional sleep health score to better predict death based on sleep, with significant associations between sleep composite score and all-cause mortality. Integrating questionnaire information and sleep microstructures, our sleep composite score is more appropriately applied for mortality risk stratification.

Physical performance strongly predicts all-cause mortality risk in a real-world population of older diabetic patients: machine learning approach for mortality risk stratification.

Prognostic risk stratification in older adults with type 2 diabetes (T2D) is important for guiding decisions concerning advance care planning. A retrospective longitudinal study was conducted in a real-world sample of older diabetic patients afferent to the outpatient facilities of the Diabetology Unit of the IRCCS INRCA Hospital of Ancona (Italy). A total of 1,001 T2D patients aged more than 70 years were consecutively evaluated by a multidimensional geriatric assessment, including physical performance evaluated using the Short Physical Performance Battery (SPPB). The mortality was assessed during a 5-year follow-up. We used the automatic machine-learning (AutoML) JADBio platform to identify parsimonious mathematical models for risk stratification. Of 977 subjects included in the T2D cohort, the mean age was 76.5 (SD: 4.5) years and 454 (46.5%) were men. The mean follow-up time was 53.3 (SD:15.8) months, and 209 (21.4%) patients died by the end of the follow-up. The JADBio AutoML final model included age, sex, SPPB, chronic kidney disease, myocardial ischemia, peripheral artery disease, neuropathy, and myocardial infarction. The bootstrap-corrected concordance index (c-index) for the final model was 0.726 (95% CI: 0.687-0.763) with SPPB ranked as the most important predictor. Based on the penalized Cox regression model, the risk of death per unit of time for a subject with an SPPB score lower than five points was 3.35 times that for a subject with a score higher than eight points (P-value <0.001). Assessment of physical performance needs to be implemented in clinical practice for risk stratification of T2D older patients.

Association of easy albumin-bilirubin score with increased mortality in adult trauma patients.

The easy albumin-bilirubin (EZ-ALBI) score is calculated using the equation: total bilirubin (mg/dl) - 9 × albumin (g/dl), and is used to evaluate liver functional reserve. This study was designed to investigate whether the EZ-ALBI score serves as an independent risk factor for mortality and is useful for stratifying the mortality risk in adult trauma patients. We retrospectively reviewed data from the registered trauma database of the hospital and included 3,637 adult trauma patients (1,241 deaths and 2,396 survivors) due to all trauma caused between January 1, 2009, and December 31, 2021. The patients were allocated to the two study groups based on the best EZ-ALBI cutoff point (EZ-ALBI = -28.5), which was determined based on the area under the receiver operating characteristic curve. Results revealed that the non-survivors had a significantly higher EZ-ALBI score than the survivors (-26.4 ± 6.5 vs. -31.5 ± 6.2, p < 0.001). Multivariate logistic regression analysis revealed that EZ-ALBI ≥ -28.5was an independent risk factor for mortality (odds ratio, 2.31; 95% confidence interval, 1.63-3.28; p < 0.001). Patients with an EZ-ALBI score ≥ -28.5 presented with 2.47-fold higher adjusted mortality rates than patients with an EZ-ALBI score < -28.5. A propensity score-matched pair cohort of 1,236 patients was developed to reduce baseline disparities in trauma mechanisms. The analysis showed that patients with an EZ-ALBI score ≥ -28.5 had a 4.12 times higher mortality rate compared to patients with an EZ-ALBI score < -28.5. The EZ-ALBI score was a significant independent risk factor for mortality and can serve as a valuable tool for stratifying mortality risk in adult trauma patients by all trauma causes.

Association between the shock index on admission and in-hospital mortality in the cardiac intensive care unit.

An elevated shock index (SI) predicts worse outcomes in multiple clinical arenas. We aimed to determine whether the SI can aid in mortality risk stratification in unselected cardiac intensive care unit patients. We included admissions to the Mayo Clinic from 2007 to 2015 and stratified them based on admission SI. The primary outcome was in-hospital mortality, and predictors of in-hospital mortality were analyzed using multivariable logistic regression. We included 9,939 unique cardiac intensive care unit patients with available data for SI. Patients were grouped by SI as follows: < 0.6, 3,973 (40%); 0.6-0.99, 4,810 (48%); and ≥ 1.0, 1,156 (12%). After multivariable adjustment, both heart rate (adjusted OR 1.06 per 10 beats per minute higher; CI 1.02-1.10; p-value 0.005) and systolic blood pressure (adjusted OR 0.94 per 10 mmHg higher; CI 0.90-0.97; p-value < 0.001) remained associated with higher in-hospital mortality. As SI increased there was an incremental increase in in-hospital mortality (adjusted OR 1.07 per 0.1 beats per minute/mmHg higher, CI 1.04-1.10, p-Value < 0.001). A higher SI was associated with increased mortality across all examined admission diagnoses. The SI is a simple and universally available bedside marker that can be used at the time of admission to predict in-hospital mortality in cardiac intensive care unit patients.

Autonomic Function and Baroreflex Control in COVID-19 Patients Admitted to the Intensive Care Unit.

Background: Autonomic function and baroreflex control might influence the survival rate of coronavirus disease 2019 (COVID-19) patients admitted to the intensive care unit (ICU) compared to respiratory failure patients without COVID-19 (non-COVID-19). This study describes physiological control mechanisms in critically ill COVID-19 patients admitted to the ICU in comparison to non-COVID-19 individuals with the aim of improving stratification of mortality risk. Methods: We evaluated autonomic and baroreflex control markers extracted from heart period (HP) and systolic arterial pressure (SAP) variability acquired at rest in the supine position (REST) and during a modified head-up tilt (MHUT) in 17 COVID-19 patients (age: 63 ± 10 years, 14 men) and 33 non-COVID-19 patients (age: 60 ± 12 years, 23 men) during their ICU stays. Patients were categorized as survivors (SURVs) or non-survivors (non-SURVs). Results: We found that COVID-19 and non-COVID-19 populations exhibited similar vagal and sympathetic control markers; however, non-COVID-19 individuals featured a smaller baroreflex sensitivity and an unexpected reduction in the HP-SAP association during the MHUT compared to the COVID-19 group. Nevertheless, none of the markers of the autonomic and baroreflex functions could distinguish SURVs from non-SURVs in either population. Conclusions: We concluded that COVID-19 patients exhibited a more preserved baroreflex control compared to non-COVID-19 individuals, even though this information is ineffective in stratifying mortality risk.

Lipoprotein Insulin Resistance Score and Mortality Risk Stratification in Heart Failure

BackgroundHigher total serum cholesterol is associated with lower mortality in heart failure. Evaluating associations between lipoprotein subfractions and mortality among people with heart failure may provide insights into this observation. MethodsWe prospectively enrolled a community cohort of people with heart failure from 2003 to 2012 and assessed vital status through 2021. Plasma collected at enrollment was used to measure lipoprotein subfractions via nuclear magnetic resonance spectroscopy. A composite score of 6 lipoprotein subfractions was generated using the lipoprotein insulin resistance index (LP-IR) algorithm. Using covariate-adjusted proportional hazards regression models, we evaluated associations between LP-IR score and all-cause mortality. ResultsAmong 1382 patients with heart failure (median follow-up 13.9 years), a one-standard-deviation (SD) increment in LP-IR score was associated with lower mortality (hazard ratio [HR] 0.93; 95% confidence interval [CI], 0.97-0.99). Among LP-IR parameters, mean high-density lipoprotein (HDL) particle size was significantly associated with lower mortality (HR per 1-SD decrement in mean HDL particle size = 0.83; 95% CI, 0.78-0.89), suggesting that the inverse association between LP-IR score and mortality may be driven by smaller mean HDL particle size. ConclusionsLP-IR score was inversely associated with mortality among patients with heart failure and may be driven by smaller HDL particle size.

TAPSE/SPAP ratio stratifies mortality risk in mild-to-moderate idiopathic pulmonary fibrosis.

<sec id="st1"><title>BACKGROUND</title>Due to paucity of literature data, we aimed at evaluating the prognostic role of the ratio of tricuspid annular plane systolic excursion (TAPSE) to systolic pulmonary artery pressure (SPAP) in idiopathic pulmonary fibrosis (IPF) patients without severe pulmonary hypertension and at assessing its correlation with effective arterial elastance index (EaI).</sec><sec id="st2"><title>METHODS</title>Multi-instrumental data obtained in 60 IPF patients (73.2 ± 6.8 years) and 60 matched controls were retrospectively analysed. Primary endpoint was all-cause mortality, while secondary endpoint was the composite of all-cause mortality and re-hospitalisations for all-causes over medium-term follow-up.</sec><sec id="st3"><title>RESULTS</title>At baseline, TAPSE/SPAP was significantly lower in patients with IPF than in controls (0.36 ± 0.25 vs. 0.77 ± 0.18 mm/mmHg; P < 0.001). TAPSE/SPAP was inversely correlated with EaI (r = -0.96) in IPF patients. During follow-up (3.5 ± 1.5 years), 21 patients died and 25 were re-hospitalised due to cardiopulmonary causes. TAPSE/SPAP was independently associated with both primary (HR 0.79, 95%CI 0.65-0.97) and secondary (HR 0.94, 95%CI 0.92-0.97) endpoints. A TAPSE/SPAP ratio of <0.20 and <0.44 mm/mmHg showed the greatest sensitivity and specificity for predicting primary (AUC 0.98) and secondary (AUC 0.99) endpoints, respectively.</sec><sec id="st4"><title>CONCLUSIONS</title>TAPSE/SPAP is a strong predictor of adverse outcomes in mild-to-moderate IPF. The strong correlation between TAPSE/SPAP and EaI might be an expression of a systemic fibrotic process which involves the heart, lungs and circulation.</sec>.

Association of preoperative and recurrent serum carcinoembryonic antigen and outcome of colorectal cancer patients with metastatic relapse

BackgroundSeldom have the associations of preoperative CEA (p-CEA) and recurrent CEA (r-CEA) levels as well as changes in p-CEA and r-CEA with survival in patients with stage I–III colorectal cancer (CRC) who have experienced metastatic relapse, been thoroughly examined. Methods241 consecutive patients with stage I–III CRC who experienced metastatic relapse at Fudan University Shanghai Cancer Center (FUSCC) between January 2008 and January 2016 were investigated. The influence of p-CEA, r-CEA and CEA alteration on the overall survival (OS) and relapse-to-death survival (RDS) was evaluated. The restricted cubic spline regression model was employed to explore the optimal cut-off value of CEA. ResultsAll 241 patients were categorized into four groups built on their CEA alteration patterns as follows: A, patients presenting elevated p-CEA levels but normal r-CEA levels (P–N); B, patients displaying normal levels of both p-CEA and r-CEA (N–N); C, patients exhibiting elevated levels of both p-CEA and r-CEA (P–P); D, patients with normal p-CEA levels but elevated r-CEA levels (N–P). The correlation between p-CEA and OS (P = 0.3266) and RDS (P = 0.2263) was insignificant. However, r-CEA exhibited a significant association with both OS (P = 0.0005) and RDS (P = 0.0002). Group A demonstrated the longest OS and RDS, whereas group D exhibited the poorest OS and RDS outcomes. For both OS and RDS, the CEA alteration groups served as an independent prognostic indicator. The optimal cut-off threshold for CEA was determined to be 5.1 ng/ml via the restricted cubic spline regression model. Conclusionr-CEA has a stronger correlation with OS and RDS in individuals with stage I–III CRC who have experienced metastatic relapse.The change between p-CEA and r-CEA could further indicate post-relapse survival, thereby facilitating the assessment of mortality risk stratification in stage I-III CRC patients experiencing metastatic relapse.

Assessing quality of life in pulmonary arterial hypertension: An independent prognostic marker.

Pulmonary arterial hypertension (PAH, or PH Group 1), a disease of aberrant pulmonary vascular remodeling, causing progressive right heart failure (RHF) due to elevation of pulmonary vascular resistance (PVR). Patient mortality risk stratification guides choice and intensity of pharmacological intervention and is assessed by haemodynamics (especially PVR) as well as noninvasive tools including WHO functional class (FC), 6-min walk distance (6MWD), and NT-proBNP levels. Quality of life (QOL) assessment is acknowledged as a central aspect of patient-centered care, but our study sought to extend QOL's role as an additional noninvasive risk marker that could further refine risk stratification and hence therapeutic choices within a "treatment to target" paradigm (aiming to achieve low-risk status). This study found that QOL assessment using the PAH-SYMPACT© physical activity tool provided enhanced, independent mortality risk information, with one unit rise in this score associated with a 41% increase in likelihood risk (odds ratio 1.41, 95% confidence interval: 1.01-1.98 (p < 0.05)) of falling within intermediate versus low-group category. We therefore found further support for additional prognostic value being conferred by measurement of QOL as part of routine PAH evaluation, reinforcing its critical role.

Radiological Parameters for the Detection of Pulmonary Hypertension in Severe Aortic Valve Stenosis and Their Influence on Mortality: Does Sex Matter?

Background: Echocardiography has long been established as the primary noninvasive method for diagnosing pulmonary hypertension (PH) prior to transcatheter aortic valve replacement (TAVR) in patients with severe aortic valve stenosis (AS). In recent years, radiological methods for diagnosing PH have been investigated. Measurements such as the computed tomography angiography (CTA)-derived pulmonary artery (PA) diameter and PA diameter/body surface area (PA/BSA) have shown promising results regarding their diagnostic strength. However, it has yet to be determined if a patient's sex has any impact on the effectiveness of these diagnostic measurements. Methods: In all, 271 patients (51.3% male, mean age 82.6 ± 4.8 years) with severe AS undergoing TAVR were separated into male and female groups. The cut-off values for the diagnosis of PH were calculated for the CTA-derived PA diameter and PA/BSA based on different systolic pulmonal artery pressure values (40-45-50 mmHg). Patients were then subclassified according to measurements above or below these PA diameters and PA/BSA cut-off values. A PA diameter ≥29.5 mm and PA/BSA ≥ 15.7 mm/m2 qualified for PH. The 1-5 year survival rate in these cohorts was further analyzed. Results: Patients with a PA diameter ≥29.5 mm showed a significantly higher 1 year mortality rate (p = 0.014). This observation could only be confirmed for the male sex (p = 0.018) and not for the female sex (p = 0.492). As for the PA/BSA, in patients over the cut-off value, no significant increase in mortality was noted in the overall cohort. However, the male patients showed increased 3 year (p = 0.048) and 5 year mortality rates (p = 0.033). Conclusions: The CTA-obtained PA diameter and PA/BSA are both useful in the diagnosis of PH and mortality risk stratification in patients with severe AS undergoing TAVR, especially in males. Male patients with PA ≥ 29.5 mm or PA/BSA ≥ 15.7 mm/m2 seem to be at a higher risk of death during follow-up after undergoing TAVR. In females, no such correlation was observed.

The CRP troponin test (CTT) stratifies mortality risk in patients with non-ST elevation myocardial infarction (NSTEMI).

The C-reactive protein (CRP)-troponin-test (CTT) comprises simultaneous serial measurements of CRP and cardiac troponin and might reflect the systemic inflammatory response in patients with acute coronary syndrome. We sought to test its ability to stratify the short- and long-term mortality risk in patients with non-ST elevation myocardial infarction (NSTEMI). We examined 1,675 patients diagnosed with NSTEMI on discharge who had at least two successive measurements of combined CRP and cardiac troponin within 48 h of admission. A tree classifier model determined which measurements and cutoffs could be used to best predict mortality during a median follow-up of 3 years [IQR 1.8-4.3]. Patients with high CRP levels ( > 90th percentile, >54 mg/L) had a higher 30-day mortality rate regardless of their troponin test findings (16.7% vs. 2.9%, p < 0.01). However, among patients with "normal" CRP levels ( < 54 mg/L), those who had high troponin levels ( > 80th percentile, 4,918 ng/L) had a higher 30-day mortality rate than patients with normal CRP and troponin concentrations (7% vs. 2%, p < 0.01). The CTT test result was an independent predictor for overall mortality even after adjusting for age, sex, and comorbidities (HR = 2.28 [95% CI 1.56-3.37], p < 0.01 for patients with high troponin and high CRP levels). Early serial CTT results may stratify mortality risk in patients with NSTEMI, especially those with "normal" CRP levels. The CTT could potentially assess the impact of inflammation during myocardial necrosis on the outcomes of patients with NSTEMI and identify patients who could benefit from novel anti-inflammatory therapies.

Comparative evaluation of non-invasive tests for risk stratification for cause specific mortality in at-risk population of hepatic fibrosis

Our study aimed to conduct a comparative evaluation of various noninvasive tests (NITs) for risk stratification in at-risk population for non-alcoholic fatty liver disease (NAFLD), focusing on cardiovascular and liver-related mortality. A total of 21,715 adults aged 40 years and older were enrolled at baseline. The mean follow-up period was 12.39 years. Three types of NITs (fibrosis-4 index [FIB-4], NAFLD fibrosis score [NFS], and steatosis-associated fibrosis estimator [SAFE] score) were used. When using the low cut-off as a 'rule-out' strategy, there were no significant differences in cardiovascular mortality between the 'rule-out' (low-risk) group and the 'rule-in' (intermediate- or high-risk) group based on FIB-4 (aHR = 1.029, P = 0.845) or NFS (aHR = 0.839, P = 0.271) classification. However, the SAFE score exhibited higher sensitivity in predicting cardiovascular mortality compared to FIB-4 or NFS (73.3% in SAFE score vs. 29.6% in FIB-4 or 21.3% in NFS). Only the SAFE score could effectively differentiate the risk between low- and intermediate- or high-risk groups for all types of mortality (all P values for aHR < 0.001). The low cutoff value of the SAFE score discriminated not only liver-related mortality but also identified the cardiovascular high-risk group in the community cohort.

Incremental Value of a Metabolic Risk Score for Heart Failure Mortality: A Population-Based Study.

Heart failure is heterogeneous syndrome with persistently high mortality. Nuclear magnetic resonance spectroscopy enables high-throughput metabolomics, suitable for precision phenotyping. We aimed to use targeted metabolomics to derive a metabolic risk score (MRS) that improved mortality risk stratification in heart failure. Nuclear magnetic resonance was used to measure 21 metabolites (lipoprotein subspecies, branched-chain amino acids, alanine, GlycA (glycoprotein acetylation), ketone bodies, glucose, and citrate) in plasma collected from a heart failure community cohort. The MRS was derived using least absolute shrinkage and selection operator penalized Cox regression and temporal validation. The association between the MRS and mortality and whether risk stratification was improved over the Meta-Analysis Global Group in Chronic Heart Failure clinical risk score and NT-proBNP (N-terminal pro-B-type natriuretic peptide) levels were assessed. The study included 1382 patients (median age, 78 years, 52% men, 43% reduced ejection fraction) with a 5-year survival rate of 48% (95% CI, 46%-51%). The MRS included 9 metabolites measured. In the validation data set, a 1 standard deviation increase in the MRS was associated with a large increased rate of death (hazard ratio, 2.2 [95% CI, 1.9-2.5]) that remained after adjustment for Meta-Analysis Global Group in Chronic Heart Failure score and NT-proBNP (hazard ratio, 1.6 [95% CI, 1.3-1.9]). These associations did not differ by ejection fraction. The integrated discrimination and net reclassification indices, and Uno's C statistic, indicated that the addition of the MRS improved discrimination over Meta-Analysis Global Group in Chronic Heart Failure and NT-proBNP. This MRS developed in a heart failure community cohort was associated with a large excess risk of death and improved risk stratification beyond an established risk score and clinical markers.

A hybrid modeling framework for generalizable and interpretable predictions of ICU mortality across multiple hospitals

The development of reliable mortality risk stratification models is an active research area in computational healthcare. Mortality risk stratification provides a standard to assist physicians in evaluating a patient’s condition or prognosis objectively. Particular interest lies in methods that are transparent to clinical interpretation and that retain predictive power once validated across diverse datasets they were not trained on. This study addresses the challenge of consolidating numerous ICD codes for predictive modeling of ICU mortality, employing a hybrid modeling approach that integrates mechanistic, clinical knowledge with mathematical and machine learning models . A tree-structured network connecting independent modules that carry clinical meaning is implemented for interpretability. Our training strategy utilizes graph-theoretic methods for data analysis, aiming to identify the functions of individual black-box modules within the tree-structured network by harnessing solutions from specific max-cut problems. The trained model is then validated on external datasets from different hospitals, demonstrating successful generalization capabilities, particularly in binary-feature datasets where label assessment involves extrapolation.

Cardiogenic Shock Prior to Percutaneous Coronary Intervention in ST-Elevation Myocardial Infarction: Outcomes and Predictors of Mortality (ANZACS-QI 73)

Cardiogenic shock (CS) is a serious complication of acute myocardial infarction (MI) and is associated with significant mortality. We describe a contemporary, real-world cohort of patients with ST-elevation MI (STEMI) and CS, including 30-day mortality and clinically relevant predictors of mortality. All patients presenting with STEMI who were treated with percutaneous coronary intervention (PCI) in New Zealand (2016 to 2020) were identified from the Aotearoa New Zealand All Cardiology Services Quality Improvement (ANZACS-QI) registry and stratified based on their Killip class on arrival to the cardiac catheterisation laboratory. Primary outcome was 30-day all-cause mortality. Multivariable analysis was used to identify predictors of mortality prior to PCI and to develop a mortality scoring system. In total, 6,649 patients were identified, including 192 (2.9%) Killip IV (CS) patients. Thirty-day mortality was 47.5% in patients with CS, 14.6% in those with heart failure without shock, and 3% in those without heart failure. Independent predictors of 30-day mortality for patients with CS were: estimated glomerular filtration rate <60 mL/min/1.73m2 (relative risk [RR] 1.89, 95% confidence interval [CI] 1.39-2.58), cardiac arrest (RR 1.54, 95% CI 1.15-2.06), diabetes (RR 1.31, 95% CI 1.01-1.70), female sex (RR 1.32, 95% CI 1.01-1.72), femoral arterial access (RR 1.42, 95% CI 1.06-1.90) and left main stem culprit (RR 2.16, 95% CI 1.65-2.84). A multivariable Shock score was developed which predicts 30-day mortality with good global discrimination (area under the curve 0.79, 95% CI 0.73-0.85). In this national cohort, the 30-day mortality for STEMI patients presenting with CS treated with PCI remains high, at nearly 50%. The ANZACS-QI Shock score is a promising tool for mortality risk stratification prior to PCI but requires further validation.

Systemic reserve dysfunction and contrast-associated acute kidney injury following percutaneous coronary intervention.

Developing contrast-associated acute kidney injury (CA-AKI) following percutaneous coronary intervention (PCI) is closely related to patient-related risk factors as well as contrast administration. The diagnostic and prognostic roles of neutrophil gelatinase-associated lipocalin (NGAL) in CA-AKI following PCI are not well established. Consecutive patients undergoing PCI were enrolled prospectively. CA-AKI was defined as an increase in the serum creatinine level ≥0.3 mg/dL within 48 hours or ≥1.5 times the baseline within 7 days after PCI. Serum NGAL concentrations were determined immediately before and 6 hours after PCI. The participants were classified into four NGAL groups according to the pre- and post-PCI NGAL values at 75th percentile. CA-AKI occurred in 38 (6.4%) of 590 patients. With chronic kidney disease status (hazard ratio [HR] 1.63, 95% confidence interval [CI]: 1.06-2.52), NGAL groups defined by the combination of pre- and 6 h post-PCI values were independently associated with the occurrence of CA-AKI (HR 1.69, 95% CI: 1.16-2.45). All-cause mortality for 29-month follow-ups was different among NGAL groups (log-rank p<0.001). Pre-PCI NGAL levels significantly correlated with baseline cardiac, inflammatory, and renal markers. Although post-PCI NGAL levels increased in patients with larger contrast administration, contrast media made a relatively limited contribution to the development of CA-AKI. In patients undergoing PCI, the combination of pre- and post-PCI NGAL values may be a useful adjunct to current risk-stratification of CA-AKI and long-term mortality. CA-AKI is likely caused by systemic reserve deficiency rather than contrast administration itself.

Intraoperative Predictors and Proposal for a Novel Prognostic Risk Score for In-Hospital Mortality after Open Repair of Ruptured Abdominal Aortic Aneurysms (SPARTAN Score).

(1) Background: Several mortality risk scores have been developed to predict mortality in ruptured abdominal aortic aneurysms (rAAAs), but none focused on intraoperative factors. The aim of this study is to identify intraoperative variables affecting in-hospital mortality after open repair and develop a novel prognostic risk score. (2) Methods: The analysis of a retrospectively maintained dataset identified patients who underwent open repair for rAAA from January 2007 to October 2023 in three Italian tertiary referral centers. Multinomial logistic regression was used to calculate the association between intraoperative variables and perioperative mortality. Independent intraoperative factors were used to create a prognostic score. (3) Results: In total, 316 patients with a mean age of 77.3 (SD ± 8.5) were included. In-hospital mortality rate was 30.7%. Hemoperitoneum (p < 0.001), suprarenal clamping (p = 0.001), and operation times of >240 min (p = 0.008) were negative predictors of perioperative mortality, while the patency of at least one hypogastric artery had a protective role (p = 0.008). Numerical values were assigned to each variable based on the respective odds ratio to create a risk stratification for in-hospital mortality. (4) Conclusions: rAAA represents a major cause of mortality. Intraoperative variables are essential to estimate patients' risk in surgically treated patients. A prognostic risk score based on these factors alone may be useful to predict in-hospital mortality after open repair.