Accurate Predictor Of Mortality Research Articles

Machine learning-based prediction of one-year mortality in ischemic stroke patients

Abstract Background Accurate prediction of mortality following an ischemic stroke is essential for tailoring personalized treatment strategies. This study evaluates the effectiveness of machine learning models in predicting one-year mortality after an ischemic stroke. Methods Five machine learning models were trained using data from a national stroke registry, with logistic regression demonstrating the highest performance. The SHapley Additive exPlanations (SHAP) analysis explained the model’s outcomes and defined the influential predictive factors. Results Analyzing 8183 ischemic stroke patients, logistic regression achieved 83% accuracy, 0.89 AUC, and an F1 score of 0.83. Significant predictors included stroke severity, pre-stroke functional status, age, hospital-acquired pneumonia, ischemic stroke subtype, tobacco use, and co-existing diabetes mellitus (DM). Discussion The model highlights the importance of predicting mortality in enhancing personalized stroke care. Apart from pneumonia, all predictors can serve the early prediction of mortality risk which supports the initiation of early preventive measures and in setting realistic expectations of disease outcomes for all stakeholders. The identified tobacco paradox warrants further investigation. Conclusion This study offers a promising tool for early prediction of stroke mortality and for advancing personalized stroke care. It emphasizes the need for prospective studies to validate these findings in diverse clinical settings.

Neutrophil-to-lymphocyte ratio as a prognostic indicator of mortality in Polycythemia Vera: insights from a prospective cohort analysis

We analyzed the neutrophil-to-lymphocyte ratio (NLR) in 1508 patients with PV and found that those with an NLR ≥ 5 were generally older, had a longer disease history, and had higher cardiovascular risk factors, more arterial thrombosis, and more aggressive blood counts, indicating a more proliferative disease. NLR was an accurate predictor of mortality, with patients with NLR ≥ 5 having significantly worse overall survival and more than twice the mortality rate compared to those with NLR < 5. Multivariable models confirmed that increasing age, previous venous thrombosis and NLR ≥ 5 were strong predictors of death, further influenced by cardiovascular risk factors. We examined the interaction between NLR and the number of cardiovascular risk factors and found a progressive trend of increased mortality risk for NLR values ≥ 5 in addition to the presence of more than one risk factor. In conclusion, patients with NLR ≥ 5 require careful monitoring and management of cardiovascular risk factors because they increase mortality when associated with progressive levels of NLR.

Prediction of Severe Acute Pancreatitis at a Very Early Stage of the Disease Using Artificial Intelligence Techniques, Without Laboratory Data or Imaging Tests: The PANCREATIA Study.

To evaluate machine learning models' performance in predicting acute pancreatitis severity using early-stage variables while excluding laboratory and imaging tests. Severe acute pancreatitis (SAP) affects approximately 20% of acute pancreatitis (AP) patients and is associated with high mortality rates. Accurate early prediction of SAP and in-hospital mortality is crucial for effective management. Traditional scores such as APACHE-II and BISAP are complex and require laboratory tests, while early predictive models are lacking. Machine learning (ML) has shown promising results in predictive modelling, potentially outperforming traditional methods. We analysed data from a prospective database of AP patients admitted to Vall d'Hebron Hospital from November 2015 to January 2022. Inclusion criteria were adults diagnosed with AP according to the 2012 Atlanta classification. Data included basal characteristics, current medication, and vital signs. We developed machine learning models to predict SAP, in-hospital mortality, and intensive care unit (ICU) admission. The modelling process included two stages: Stage 0, which used basal characteristics and medication, and Stage 1, which included data from Stage 0 and vital signs. Out of 634 cases, 594 were analysed. The Stage 0 model showed AUC values of 0.698 for mortality, 0.721 for ICU admission, and 0.707 for persistent organ failure. The Stage 1 model improved performance with AUC values of 0.849 for mortality, 0.786 for ICU admission, and 0.783 for persistent organ failure. The models demonstrated comparable or superior performance to APACHE-II and BISAP scores. The ML models showed good predictive capacity for SAP, ICU admission, and mortality using early-stage data without laboratory or imaging tests. This approach could revolutionise AP patients' initial triage and management, providing a personalised prediction method based on early clinical data.

Adjustment of the SMART risk score by bioactive adrenomedullin enables a more accurate prediction of mortality in patients with ASCVD

Abstract Background Bioactive adrenomedullin 1-52 (bio-ADM) is a dynamic blood biomarker for real-time assessment of endothelial function. Bio-ADM was recently shown to be a prognostic marker in patients with acute heart failure and cardiogenic shock. The SMART (Second Manifestations of Arterial Disease) score is a validated tool for risk assessment in patients with established atherosclerotic cardiovascular disease (ASCVD). The aim of this study was to assess whether measurement of bio-ADM adds incremental value to the SMART score in stable patients with ASCVD. Methods Circulating bio-ADM levels were assessed in n=695 stable patients with ASCVD in an all-comer cohort. Endpoints evaluated were all-cause mortality and cardiovascular mortality; follow up was 3 years. Results Bio-ADM was higher in non-survivors (all-cause death: n=54, median 33.1 pg/mL) compared to survivors (n=641, median 17.9 pg/mL; p&amp;lt;0.0001). Univariable Cox regression analyses showed bio-ADM to be associated with adverse outcome [standardized hazard ratio (HR) of bio-ADM values: All-cause death: 2.4, 95% confidence interval (CI): 2.0-2.9; p&amp;lt;0.001, cardiovascular death: 2.5, 95% CI: 1.9-3.3; p&amp;lt;0.001]. This association remained significant in various multivariable Cox regression models. Bio-ADM was found to be a strong marker for mortality (c-index: 0.80, Chi²: 54.1) and proved to be superior to other markers including hs-Troponin T (c-index: 0.61, Chi²: 2.0) and eGFR CKD-EPI 2021 (c-index: 0.687, Chi²: 33.5). Addition of bio-ADM to the SMART score significantly improved model performance in predicting mortality (SMART score: c-index: 0.717, Chi²: 24.73; SMART score + bio-ADM: c-index: 0.832, Chi²: 63.24; Delta c-index: 0.115; Delta Chi²: 38.51; all p&amp;lt;0.001) and showed net reclassification improvement in risk categorization with 11.6% in the group of survivors and 9.8% for non-survivors. Conclusion Bio-ADM provides incremental added value (improved discrimination, calibration and reclassification) on top of the SMART risk score in patients with ASCVD.

Prediction of in-hospital mortality risk for patients with acute ST-elevation myocardial infarction after primary PCI based on predictors selected by GRACE score and two feature selection methods.

Accurate in-hospital mortality prediction following percutaneous coronary intervention (PCI) is crucial for clinical decision-making. Machine Learning (ML) and Data Mining methods have shown promise in improving medical prognosis accuracy. We analyzed a dataset of 4,677 patients from the Regional Vascular Center of Primorsky Regional Clinical Hospital No. 1 in Vladivostok, collected between 2015 and 2021. We utilized Extreme Gradient Boosting, Histogram Gradient Boosting, Light Gradient Boosting, and Stochastic Gradient Boosting for mortality risk prediction after primary PCI in patients with acute ST-elevation myocardial infarction. Model selection was performed using Monte Carlo Cross-validation. Feature selection was enhanced through Recursive Feature Elimination (RFE) and Shapley Additive Explanations (SHAP). We further developed hybrid models using Augmented Grey Wolf Optimizer (AGWO), Bald Eagle Search Optimization (BES), Golden Jackal Optimizer (GJO), and Puma Optimizer (PO), integrating features selected by these methods with the traditional GRACE score. The hybrid models demonstrated superior prediction accuracy. In scenario (1), utilizing GRACE scale features, the Light Gradient Boosting Machine (LGBM) and Extreme Gradient Boosting (XGB) models optimized with BES achieved Recall values of 0.944 and 0.954, respectively. In scenarios (2) and (3), employing SHAP and RFE-selected features, the LGB models attained Recall values of 0.963 and 0.977, while the XGB models achieved 0.978 and 0.99. The study indicates that ML models, particularly the XGB optimized with BES, can outperform the conventional GRACE score in predicting in-hospital mortality. The hybrid models' enhanced accuracy presents a significant step forward in risk assessment for patients post-PCI, offering a potential alternative to existing clinical tools. These findings underscore the potential of ML in optimizing patient care and outcomes in cardiovascular medicine.

Development and validation of a chronic kidney disease progression model using patient-level simulations

Chronic disease progression models are available for several highly prevalent conditions. For chronic kidney disease (CKD), the scope of existing progression models is limited to the risk of kidney failure and major cardiovascular (CV) events. The aim of this project was to develop a comprehensive CKD progression model (CKD-PM) that simulates the risk of CKD progression and a broad range of complications in patients with CKD. A series of literature reviews informed the selection of risk factors and identified existing risk equations/algorithms for kidney replacement therapy (KRT), CV events, other CKD-related complications, and mortality. Risk equations and transition probabilities were primarily sourced from publications produced by large US and international CKD registries. A patient-level, state-transition model was developed with health states defined by the Kidney Disease Improving Global Outcomes categories. Model validation was performed by comparing predicted outcomes with observed outcomes in the source cohorts used in model development (internal validation) and other cohorts (external validation). The CKD-PM demonstrated satisfactory modeling properties. Accurate prediction of all-cause and CV mortality was achieved without calibration, while prediction of CV events through CKD-specific equations required implementation of a calibration factor to balance time-dependent versus baseline risk. Predicted annual changes in estimated glomerular filtration rate (eGFR) and urine albumin-creatinine ratio were acceptable in comparison to external values. A flexible eGFR threshold for KRT equations enabled accurate prediction of these events. This CKD-PM demonstrated reliable modeling properties. Both internal and external validation revealed robust outcomes.

Comprehensive risk factor-based nomogram for predicting one-year mortality in patients with sepsis-associated encephalopathy

Sepsis-associated encephalopathy (SAE) is a frequent and severe complication in septic patients, characterized by diffuse brain dysfunction resulting from systemic inflammation. Accurate prediction of long-term mortality in these patients is critical for improving clinical outcomes and guiding treatment strategies. We conducted a retrospective cohort study using the MIMIC IV database to identify adult patients diagnosed with SAE. Patients were randomly divided into a training set (70%) and a validation set (30%). Least absolute shrinkage and selection operator regression and multivariate logistic regression were employed to identify significant predictors of 1-year mortality, which were then used to develop a prognostic nomogram. The model’s discrimination, calibration, and clinical utility were assessed using the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis, respectively. A total of 3,882 SAE patients were included in the analysis. The nomogram demonstrated strong predictive performance with AUCs of 0.881 (95% CI: 0.865, 0.896) in the training set and 0.859 (95% CI: 0.830, 0.888) in the validation set. Calibration plots indicated good agreement between predicted and observed 1-year mortality rates. The decision curve analysis showed that the nomogram provided greater net benefit across a range of threshold probabilities compared to traditional scoring systems such as Glasgow Coma Scale and Sequential Organ Failure Assessment. Our study presents a robust and clinically applicable nomogram for predicting 1-year mortality in SAE patients. This tool offers superior predictive performance compared to existing severity scoring systems and has significant potential to enhance clinical decision-making and patient management in critical care settings.

“An Analytical Study To Compare The Effectiveness Of P-Possum And Possum Scores In Predicting Mortality And Morbidity Following Emergency Laparotomy"

Background: Surgical outcomes are increasingly scrutinized as part of efforts to ensure accountability and improve the quality of care. The Physiological and Operative Severity Score for the enumeration of Mortality and Morbidity (POSSUM) and its modification, the Portsmouth POSSUM (P-POSSUM), are widely used tools for predicting surgical outcomes. Objective: This study aimed to assess the accuracy of POSSUM in predicting morbidity and P-POSSUM in predicting mortality for patients undergoing emergency laparotomies in a North Indian tertiary care center. Methods: A hospital-based, analytic observational study was conducted at S.M.S. Medical College and Hospital, Jaipur, involving 160 patients who underwent emergency laparotomy. Preoperative physiological and intraoperative data were collected and used to calculate POSSUM and P-POSSUM scores. Mortality predictions were compared using linear and exponential analyses, and the observed-to-expected (O:E) mortality ratio was calculated. Statistical significance was determined using the Chi-square test. Results: The mean age of the cohort was 49.65 years, with males comprising 64.37% of the population. Peptic perforation (20.6%) was the most common cause of surgery. POSSUM significantly over-predicted mortality (O:E ratio = 0.26, p=0.04), while P-POSSUM provided a more accurate prediction (O:E ratio = 0.45, p=0.81). POSSUM also overestimated morbidity (O:E ratio = 0.8, p=0.51). Conclusion: While POSSUM tends to over-predict both mortality and morbidity, P-POSSUM offers a more accurate prediction of mortality in patients undergoing emergency laparotomy. These scoring systems can serve as valuable tools for surgical audits and improving the quality of care, though careful selection of the appropriate predictive model is essential

Interpretable mortality prediction model for ICU patients with pneumonia: using shapley additive explanation method

BackgroundPneumonia, a leading cause of morbidity and mortality worldwide, often necessitates Intensive Care Unit (ICU) admission. Accurate prediction of pneumonia mortality is crucial for tailored prevention and treatment plans. However, existing mortality prediction models face limited adoption in clinical practice due to their lack of interpretability.ObjectiveThis study aimed to develop an interpretable model for predicting pneumonia mortality in ICUs. Leveraging the Shapley Additive Explanation (SHAP) method, we sought to elucidate the Extreme Gradient Boosting (XGBoost) model and identify prognostic factors for pneumonia.MethodsConducted as a retrospective cohort study, we utilized electronic health records from the eICU-CRD (2014–2015) for all adult pneumonia patients. The first 24 h of each ICU admission records were considered, with 70% of the dataset allocated for model training and 30% for validation. The XGBoost model was employed, and performance was assessed using the area under the receiver operating characteristic curve (AUC). The SHAP method provided insights into the XGBoost model.ResultsAmong 10,962 pneumonia patients, in-hospital mortality was 16.33%. The XGBoost model demonstrated superior predictive performance (AUC: 0.778 ± 0.016)) compared to traditional scoring systems and other machine learning method, which achieved an improvement of 10% points. SHAP analysis identified Aspartate Aminotransferase (AST) as the most crucial predictor.ConclusionsInterpretable predictive models enhance mortality risk assessment for pneumonia patients in the ICU, fostering transparency. AST emerged as the foremost predictor, followed by patient age, albumin, BMI et al. These insights, rooted in strong correlations with mortality, facilitate improved clinical decision-making and resource allocation.

Assessing the Utility of Prediction Scores PAINT, ISARIC4C, CHIS, and COVID-GRAM at Admission and Seven Days after Symptom Onset for COVID-19 Mortality.

The COVID-19 pandemic underscores the need for accurate prognostic tools to predict patient outcomes. This study evaluates the effectiveness of four prominent COVID-19 prediction scores-PAINT, ISARIC4C, CHIS, and COVID-GRAM-at two critical time points: at admission and seven days post-symptom onset, to assess their utility in predicting mortality among hospitalized patients. Conducted at the Clinical Emergency Hospital Pius Brînzeu in Timișoara, this retrospective analysis included adult patients hospitalized with confirmed SARS-CoV-2 infection. Eligible patients had complete data for the scores at both time points. Statistical analysis involved ROC curves and logistic regression to assess the scores' predictive accuracy for mortality. The study included 215 patients, split into 139 survivors and 76 non-survivors. At admission, the PAINT, ISARIC4C, CHIS, and COVID-GRAM scores significantly differentiated between the survival outcomes (p < 0.0001). The best cutoff values at admission were 6.26 for PAINT, 7.95 for ISARIC4C, 5.58 for CHIS, and 0.63 for COVID-GRAM, corresponding to sensitivities of 85.47%, 80.56%, 88.89%, and 83.33% and specificities of 77.34%, 82.12%, 75.01%, and 78.45%, respectively. By day seven, the cutoff values increased, indicating deteriorating conditions in patients who eventually succumbed to the virus. The hazard ratios at admission for exceeding these cutoffs were significant: PAINT (HR = 3.45), ISARIC4C (HR = 2.89), CHIS (HR = 4.02), and COVID-GRAM (HR = 3.15), highlighting the scores' abilities to predict severe outcomes. One week post symptom onset, these scores' predictive values and corresponding hazard ratios increased, further validating their prognostic significance over time. The evaluated COVID-19 prediction scores robustly predict mortality at admission and become more predictive by the seventh day of symptom onset. These findings support the use of these scores in clinical settings to facilitate early identification and intervention for high-risk patients, potentially improving patient outcomes during the ongoing global health crisis.

The Naples Score: Can it Outperform Existing Scores in Predicting Gastric Cancer Mortality?

Objective: Gastric surgery, including curative and palliative procedures, is crucial in managing gastric cancer. Despite advancements, postoperative mortality remains a significant concern, influenced by factors such as patient nutritional status. Accurate assessment of nutritional status is essential for risk stratification and improving patient outcomes.This study aims to investigate the correlation between four nutritional scores: Naples Prognostic Score , Controlling Nutritional Status score, Prognostic Nutritional Index, and Systemic Inflammation Score and the mortality rate following gastric surgery, to identify the most reliable predictors of postoperative mortality. Materials and methods: This retrospective study evaluated 50 patients diagnosed with gastric adenocarcinoma and operated on at Erol Olçok Training and Research Hospital between April 2021 and September 2023. Data collected included age, gender, operation type, laparoscopy usage, albumin, cholesterol, neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, mortality rates, and TNM stages. Nutritional scores were calculated, and their predictive accuracy for mortality was assessed using time-dependent Receiver Operating Characteristic curve analysis. Results: The study included 50 patients. Significant differences in albumin levels, neutrophil-to-lymphocyte rate, lymphocyte-to-monocyte ratio, and nutritional scores were found between deceased and surviving patients. Prognostic Nutritional Index had the highest predictive accuracy with an Area under the curve of 0.792, sensitivity of 76%, and specificity of 86.2%, outperforming others. Conclusion: The Prognostic Nutritional Index is the most effective predictor of postoperative mortality among the evaluated scores.

Comparison of Qanadli score with conventional risk stratifiers in non-massive pulmonary emboli.

The management and risk stratification of non-massive pulmonary embolism (PE) remain unclear. However, early assessment of PE severity can aid physicians in establishing better treatment milestones. There has been no direct comparison of mortality rates in patients with non-massive PE, and existing data are sometimes contradictory. Therefore, we examined the relationship between the Qanadli index and conventional risk stratifiers in PE. We retrospectively analyzed 200 consecutively selected patients diagnosed with PE. The assessment included computed tomography pulmonary angiography, electrocardiography, echocardiography findings, outcomes, and a comparison with the Simplified Pulmonary Embolism Severity Index (SPESI) score. Descriptive, regression, and receiver operating characteristic analyses were performed. The mean Qanadli score was 13.5 ± 1.15. Pearson correlation analysis revealed significant associations between the total Qanadli score and several variables: right ventricular enlargement, follow-up ejection fraction, and SPESI score. Although the Qanadli score did not significantly predict mortality, the risk of death increased by 58.8% for each 1-unit increase in the SPESI score. Although the Qanadli index is valuable in assessing PE and guiding treatment strategies, its standalone predictive value for mortality may be insufficient. Therefore, incorporating scoring systems such as the SPESI and echocardiographic findings is recommended for more accurate mortality prediction.

Hematocrit and Albumin Levels at Admission Predict in-Hospital Mortality in Pediatric COVID-19 Omicron Variant Patients.

The Omicron variant is the present predominant COVID-19 strain worldwide. Accurate mortality prediction can facilitate risk stratification and targeted therapies. The study aimed to evaluate the feasibility of the difference in hematocrit and albumin (HCT-ALB) levels, alone or combined with the pediatric Sequential Organ Failure Assessment (pSOFA) score and lactate level, to predict the in-hospital mortality of COVID-19 Omicron variant-infected pediatric patients. A multicenter retrospective cohort study was performed for children with COVID-19 Omicron variant infection between December 2021 and January 2022. The demographics, clinical characteristics, hospital admission laboratory test results, and treatments were recorded. The in-hospital mortality was documented. The associations between HCT-ALB levels and mortality, and between HCT-ALB+pSOFA+lactate and mortality were analyzed. A total of 119 children were included. The median age was 1.6 (interquartile range: 0.5-6.2) years old. There were 70 boys and 49 girls. The mortality rate was 14.3% (17/119). The univariate and multivariate Cox regression analysis revealed that HCT-ALB was associated to in-hospital mortality (hazard ratio: 1.500, 95% confidence interval: 1.235-1.822, p<0.001). The receiver operating characteristic curve analysis revealed that HCT-ALB can be used to accurately predict in-hospital mortality at a cut-off value of -0.7 (area under the curve: 0.888, sensitivity: 0.882, specificity: 0.225, Youden index: 0.657, p<0.001). These patients were assigned into three groups based on the HCT-ALB level, pSOFA score, and lactate level (low-, medium-, and high-risk groups). The Kaplan-Meier analysis revealed that the mortality increased in the high-risk group, when compared to the medium-risk group (p<0.01). The latter group had a higher mortality, when compared to the low-risk group (p<0.01). The HCT-ALB level can be applied to predict the in-hospital mortality of children infected with the COVID-19 Omicron variant. Its combination with other variables can improve prediction performance.

Artificial Intelligence-Based Models for Prediction of Mortality in ICU Patients: A Scoping Review.

Healthcare professionals may be able to anticipate more accurately a patient's timing of death and assess their possibility of recovery by implementing a real-time clinical decision support system. Using such a tool, the healthcare system can better understand a patient's condition and make more informed judgements about distributing limited resources. This scoping review aimed to analyze various death prediction AI (Artificial Intelligence) algorithms that have been used in ICU (Intensive Care Unit) patient populations. The search strategy of this study involved keyword combinations of outcome and patient setting such as mortality, survival, ICU, terminal care. These terms were used to perform database searches in MEDLINE, Embase, and PubMed up to July 2022. The variables, characteristics, and performance of the identified predictive models were summarized. The accuracy of the models was compared using their Area Under the Curve (AUC) values. Databases search yielded an initial pool of 8271 articles. A two-step screening process was then applied: first, titles and abstracts were reviewed for relevance, reducing the pool to 429 articles. Next, a full-text review was conducted, further narrowing down the selection to 400 key studies. Out of 400 studies on different tools or models for prediction of mortality in ICUs, 16 papers focused on AI-based models which were ultimately included in this study that have deployed different AI-based and machine learning models to make a prediction about negative patient outcome. The accuracy and performance of the different models varied depending on the patient populations and medical conditions. It was found that AI models compared with traditional tools like SAP3 or APACHE IV score were more accurate in death prediction, with some models achieving an AUC of up to 92.9%. The overall mortality rate ranged from 5% to more than 60% in different studies. We found that AI-based models exhibit varying performance across different patient populations. To enhance the accuracy of mortality prediction, we recommend customizing models for specific patient groups and medical contexts. By doing so, healthcare professionals may more effectively assess mortality risk and tailor treatments accordingly. Additionally, incorporating additional variables-such as genetic information-into new models can further improve their accuracy.

Physical Function and Mortality in Older Adults with Chronic Kidney Disease.

Accurate mortality prediction can guide clinical care for older adults with chronic kidney disease (CKD). Yet existing tools do not incorporate physical function, an independent predictor of death in older adults. We determined whether incorporating physical function measurements improve mortality prediction among older adults with CKD. We included Chronic Renal Insufficiency Cohort participants who were ≥65 years old, had estimated Glomerular Filtration Rate (eGFR) <60 mL/min/1.73m2, not receiving kidney failure replacement therapy (KFRT), and had least one gait speed assessment. Gait speed was measured at usual pace (≥0.84, 0.83-0.65, 0.64-0.47, ≤0.46 meters/second, or unable), and frailty was assessed using Physical Frailty Phenotype criteria (range 0 to 5 points, also known as Fried criteria). We modeled time to all-cause death over five years using Cox proportional hazard models, treating KFRT as censored and non-censored events in separate analyses. C-statistics assessed model discrimination. Among 2,338 persons, mean age was 70±4 years; 43% were female and 43% were Black. Mean eGFR was 42±13 mL/min/1.73m2 and median urine albumin-to-creatinine ratio was 33 mg/g [Q1 9, Q3 206]. Over a median follow-period of 5 years, 392 died and 164 developed KFRT. In censored analyses, adding gait speed or frailty improved mortality risk prediction. The C-statistic changed from 0.69 to 0.72 with gait speed scores, and from 0.70 to 0.73 with frailty scores. The performance of models with gait speed or frailty was similar in non-censored analyses. Among older adults with CKD, adding measures of physical function modestly improves mortality prediction.

Leveraging machine learning to enhance postoperative risk assessment in coronary artery bypass grafting patients with unprotected left main disease - A retrospective cohort study.

Risk stratification for patients undergoing coronary artery bypass surgery (CABG) for left main coronary artery (LMCA) disease is essential for informed decision-making. This study explored the potential of machine learning (ML) methods to identify key risk factors associated with mortality in this patient group. This retrospective cohort study was conducted on 866 patients from the Gulf Left Main Registry who presented between 2015 and 2019. The study outcome was hospital all-cause mortality. Various machine learning models [logistic regression, random forest (RF), k-nearest neighbor, support vector machine, naïve Bayes, multilayer perception, boosting] were used to predict mortality, and their performance was measured using accuracy, precision, recall, F1 score, and area under the receiver operator characteristic curve (AUC). Nonsurvivors had significantly greater EuroSCORE II values (1.84 (10.08-3.67) vs. 4.75 (2.54-9.53) %, P<0.001 for survivors and nonsurvivors, respectively). The EuroSCORE II score significantly predicted hospital mortality (OR: 1.13 (95% confidence interval: 1.09-1.18), P<0.001), with an AUC of 0.736. RF achieved the best ML performance (accuracy=98, precision=100, recall=97 and F1 score=98). Explainable artificial intelligence using SHAP demonstrated the most important features as follows: preoperative lactate level, emergency surgery, chronic kidney disease (CKD), NSTEMI, nonsmoking status, and sex. QLattice identified lactate and CKD as the most important factors for predicting hospital mortality this patient group. This study demonstrates the potential of ML, particularly the Random Forest, to accurately predict hospital mortality in patients undergoing CABG for LMCA disease and its superiority over traditional methods. The key risk factors identified, including preoperative lactate levels, emergency surgery, chronic kidney disease, NSTEMI, nonsmoking status, and sex, provide valuable insights for risk stratification and informed decision-making in this high-risk patient population. Additionally, incorporating newly identified risk factors into future risk scoring systems can further improve mortality prediction accuracy.

Nomogram to Predict 90-Day All-Cause Mortality in Acute Ischemic Stroke Patients after Endovascular Thrombectomy.

Although Endovascular Thrombectomy (EVT) significantly improves the prognosis of Acute Ischemic Stroke (AIS) patients with large vessel occlusion, the mortality rate remains higher. This study aimed to construct and validate a nomogram for predicting 90-day all-cause mortality in AIS patients with large vessel occlusion and who have undergone EVT. AIS patients with large vessel occlusion in the anterior circulation who underwent EVT from May 2017 to December 2022 were included. 430 patients were randomly split into a training group (N=302) and a test group (N=128) for the construction and validation of our nomogram. In the training group, multivariate logistic regression analysis was performed to determine the predictors of 90-day all-cause mortality. The C-index, calibration plots, and decision curve analysis were applied to evaluate the nomogram performance. Multivariate logistic regression analysis revealed neurological deterioration during hospitalization, age, baseline National Institutes of Health Stroke Scale (NIHSS) score, occlusive vessel location, malignant brain edema, and Neutrophil-to-lymphocyte Ratio (NLR) as the independent predictors of 90-day all-cause mortality (all p ≤ 0.039). The C-index of the training and test groups was 0.891 (95%CI 0.848-0.934) and 0.916 (95% CI: 0.865-0.937), respectively, showing the nomogram to be well distinguished. The Hosmer-Lemeshow goodness-of-fit test revealed the p-values for both the internal and external verification datasets to be greater than 0.5. Our nomogram has incorporated relevant clinical and imaging features, including neurological deterioration, age, baseline NIHSS score, occlusive vessel location, malignant brain edema, and NLR ratio, to provide an accurate and reliable prediction of 90-day all-cause mortality in AIS patients undergoing EVT.

A plasma peptidomic signature reveals extracellular matrix remodeling and predicts prognosis in alcohol-associated hepatitis.

Alcohol-associated hepatitis (AH) is plagued with high mortality and difficulty in identifying at-risk patients. The extracellular matrix undergoes significant remodeling during inflammatory liver injury and could potentially be used for mortality prediction. EDTA plasma samples were collected from patients with AH (n = 62); Model for End-Stage Liver Disease score defined AH severity as moderate (12-20; n = 28) and severe (>20; n = 34). The peptidome data were collected by high resolution, high mass accuracy UPLC-MS. Univariate and multivariate analyses identified differentially abundant peptides, which were used for Gene Ontology, parent protein matrisomal composition, and protease involvement. Machine-learning methods were used to develop mortality predictors. Analysis of plasma peptides from patients with AH and healthy controls identified over 1600 significant peptide features corresponding to 130 proteins. These were enriched for extracellular matrix fragments in AH samples, likely related to the turnover of hepatic-derived proteins. Analysis of moderate versus severe AH peptidomes was dominated by changes in peptides from collagen 1A1 and fibrinogen A proteins. The dominant proteases for the AH peptidome spectrum appear to be CAPN1 and MMP12. Causal graphical modeling identified 3 peptides directly linked to 90-day mortality in >90% of the learned graphs. These peptides improved the accuracy of mortality prediction over the Model for End-Stage Liver Disease score and were used to create a clinically applicable mortality prediction assay. A signature based on plasma peptidome is a novel, noninvasive method for prognosis stratification in patients with AH. Our results could also lead to new mechanistic and/or surrogate biomarkers to identify new AH mechanisms.

Accurate prediction of all-cause mortality in patients with metabolic dysfunction-associated steatotic liver disease using electronic health records

Introduction and ObjectivesDespite the huge clinical burden of MASLD, validated tools for early risk stratification are lacking, and heterogeneous disease expression and a highly variable rate of progression to clinical outcomes result in prognostic uncertainty. We aimed to investigate longitudinal electronic health record-based outcome prediction in MASLD using a state-of-the-art machine learning model. Patients and Methodsn = 940 patients with histologically-defined MASLD were used to develop a deep-learning model for all-cause mortality prediction. Patient timelines, spanning 12 years, were fully-annotated with demographic/clinical characteristics, ICD-9 and -10 codes, blood test results, prescribing data, and secondary care activity. A Transformer neural network (TNN) was trained to output concomitant probabilities of 12-, 24-, and 36-month all-cause mortality. In-sample performance was assessed using 5-fold cross-validation. Out-of-sample performance was assessed in an independent set of n = 528 MASLD patients. ResultsIn-sample model performance achieved AUROC curve 0.74–0.90 (95 % CI: 0.72–0.94), sensitivity 64 %-82 %, specificity 75 %–92 % and Positive Predictive Value (PPV) 94 %-98 %. Out-of-sample model validation had AUROC 0.70–0.86 (95 % CI: 0.67–0.90), sensitivity 69 %–70 %, specificity 96 %–97 % and PPV 75 %–77 %. Key predictive factors, identified using coefficients of determination, were age, presence of type 2 diabetes, and history of hospital admissions with length of stay >14 days. ConclusionsA TNN, applied to routinely-collected longitudinal electronic health records, achieved good performance in prediction of 12-, 24-, and 36-month all-cause mortality in patients with MASLD. Extrapolation of our technique to population-level data will enable scalable and accurate risk stratification to identify people most likely to benefit from anticipatory health care and personalized interventions.

The performance of metabolomics-based prediction scores for mortality in older patients with solid tumors

Prognostic information is needed to balance benefits and risks of cancer treatment in older patients. Metabolomics-based scores were previously developed to predict 5- and 10-year mortality (MetaboHealth) and biological age (MetaboAge). This study aims to investigate the association of MetaboHealth and MetaboAge with 1-year mortality in older patients with solid tumors, and to study their predictive value for mortality in addition to established clinical predictors. This prospective cohort study included patients aged ≥ 70 years with a solid malignant tumor, who underwent blood sampling and a geriatric assessment before treatment initiation. The outcome was all-cause 1-year mortality. Of the 192 patients, the median age was 77 years. With each SD increase of MetaboHealth, patients had a 2.32 times increased risk of mortality (HR 2.32, 95% CI 1.59–3.39). With each year increase in MetaboAge, there was a 4% increased risk of mortality (HR 1.04, 1.01–1.07). MetaboHealth and MetaboAge showed an AUC of 0.66 (0.56–0.75) and 0.60 (0.51–0.68) for mortality prediction accuracy, respectively. The AUC of a predictive model containing age, primary tumor site, distant metastasis, comorbidity, and malnutrition was 0.76 (0.68–0.83). Addition of MetaboHealth increased AUC to 0.80 (0.74–0.87) (p = 0.09) and AUC did not change with MetaboAge (0.76 (0.69–0.83) (p = 0.89)). Higher MetaboHealth and MetaboAge scores were associated with 1-year mortality. The addition of MetaboHealth to established clinical predictors only marginally improved mortality prediction in this cohort with various types of tumors. MetaboHealth may potentially improve identification of older patients vulnerable for adverse events, but numbers were too small for definitive conclusions. The TENT study is retrospectively registered at the Netherlands Trial Register (NTR), trial number NL8107. Date of registration: 22–10-2019.