Multicenter Machine Learning Model for Assessing the Impact of Malignancy on In-Hospital Mortality in Heart Failure Patients: A Clinical Decision Support System with Interpretable Artificial Intelligence

Multicenter Machine Learning Model for Assessing the Impact of Malignancy on In-Hospital Mortality in Heart Failure Patients: A Clinical Decision Support System with Interpretable Artificial Intelligence | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Multicenter Machine Learning Model for Assessing the Impact of Malignancy on In-Hospital Mortality in Heart Failure Patients: A Clinical Decision Support System with Interpretable Artificial Intelligence Tao Zhang, Zihan Nan, Shaohan Guo, Shiju Yang, Leying Li, Yifei Zhang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7179137/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Background Heart failure (HF) and malignancy represent two major global health burdens that frequently coexist and lead to poor clinical outcomes. However, the specific impact of malignancy on in-hospital mortality in HF patients remains incompletely understood, and reliable predictive tools specifically for HF patients with comorbid malignancy are currently lacking. Methods This multicenter retrospective study analyzed data from the eICU Collaborative Research Database (eICU database), Medical Information Mart for Intensive Care IV (MIMIC-IV) databases, and Medical Information Mart for Intensive Care III (MIMIC-III) databases, including 21,636 HF patients (3,397 with malignancy). We employed three analytical approaches: propensity score matching (PSM), inverse probability treatment weighting (IPTW), and multivariable logistic regression to assess malignancy-associated mortality risk. For predictive modeling, five machine learning algorithms were trained on the eICU database (70% training, 30% internal validation) and externally validated using both MIMIC-IV and MIMIC-III datasets. Results All three analytical methods (PSM, IPTW, and multivariable regression) yielded highly consistent results, demonstrating that malignancy significantly increased in-hospital mortality risk (PSM: OR 1.14, 95% CI 1.02–1.26; IPTW: OR 1.16, 95% CI 1.03–1.30; multivariable regression: OR 1.20, 95% CI 1.07–1.35). The AdaBoost model, developed using 19 key predictive variables selected by the Boruta algorithm, demonstrated excellent performance with a training set AUC of 0.849 and internal validation AUC of 0.740, while maintaining good discriminative ability in external validation (MIMIC-IV: AUC 0.739; MIMIC-III: AUC 0.699). To enhance model interpretability, SHapley Additive exPlanations analysis revealed the top five predictive variables: Simplified Acute Physiology Score II, mechanical ventilation requirement, heart rate, body temperature, and respiratory rate. For clinical implementation, we developed a web-based calculator (available at: https://nanzihan1998.shinyapps.io/Mortality/ ) to facilitate real-time mortality risk assessment. Conclusions Malignancy independently worsens outcomes in HF patients. Our interpretable machine learning model incorporating multiple clinically relevant predictors provides accurate mortality risk stratification, facilitating personalized clinical decision-making for this high-risk population. Future studies should incorporate longitudinal data and novel biomarkers to further improve predictive performance. Health sciences/Biomarkers Biological sciences/Cancer Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Medical research Health sciences/Oncology Health sciences/Risk factors heart failure malignancy in-hospital mortality machine learning interpretable AI predictive modeling Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Heart failure (HF) and malignancy represent two of the most significant global health challenges, with both conditions showing rising prevalence due to aging populations 1 . HF affects over 56 million people worldwide and is associated with high morbidity and mortality, including a 28.2% all-cause mortality rate at 3 years post-diagnosis 2,3 . Meanwhile, malignancy remains the second leading cause of death globally, accounting for 14.57% of total mortality 4 . Notably, these conditions frequently coexist, particularly in older adults—approximately 10% of patients with malignancy aged ≥ 65 years also have HF 5 . This overlap poses a substantial clinical burden, as patients with both HF and malignancy experience worse outcomes, including prolonged hospital stays, increased healthcare costs, and nearly threefold higher in-hospital mortality compared to those with either condition alone 6 . Accurate prognostic prediction in patients with concurrent HF and malignancy is crucial for optimizing therapeutic decision-making, yet remains a significant clinical challenge. The intricate interplay between these conditions—encompassing shared risk factors, overlapping pathophysiological pathways, and malignancy treatment-related cardiotoxicity—substantially complicates risk stratification and clinical management. A comprehensive elucidation of malignancy-associated outcomes in HF populations could facilitate more personalized and effective care delivery. Recent advances in data science and artificial intelligence have facilitated the increasing application of machine learning (ML) in prognostic analysis for critically ill Intensive Care Unit (ICU) patients. Especially with the development of interpretable ML techniques, the "black box" of traditional ML has been unveiled, making the model's results more convincing. In the field of HF, interpretable ML techniques have been widely adopted. For instance, Jili Li's team employed an interpretable ML model to predict mortality risk in HF patients admitted to the ICU 7 . Similarly, Shengxian Peng et al. developed an interpretable ML approach to forecast 28-day all-cause in-hospital mortality among critically ill HF patients with comorbid hypertension 8 . Additionally, Nicklas Vinter's team created an interpretable ML model to predict the risk of new-onset atrial fibrillation in HF patients 9 . In the field of oncology, interpretable ML techniques have also demonstrated significant advancements. For example, the NKECLR model developed by Alphonse Houssou Hounye's team enables precise prediction of survival rates and treatment responses in melanoma patients 10 , while Mitra Montazeri's research group systematically analyzed survival outcomes across different breast cancer subtypes using ML algorithms 11 . However, in the emerging interdisciplinary field of cardio-oncology, prognostic prediction research specifically targeting HF patients with comorbid malignancies remains notably underdeveloped. Therefore, this study aims to: (1) systematically evaluated the malignancy-associated in-hospital mortality risk in HF patients; (2) develop a prognostic model for predicting in-hospital mortality specifically in HF patients with comorbid malignancy; and (3) conduct interpretability analysis of the model with visualization to facilitate clinical implementation. 2. Methods 2.1 Overview of the Methods This study employed a three-stage analytical approach: (1) multidimensional confounding adjustment and risk quantification using propensity score matching (PSM), inverse probability treatment weighting (IPTW), and multivariable logistic regression, confirming significantly worse clinical outcomes in HF patients with malignancies; (2) partitioning of the eICU Collaborative Research Database(eICU database) into training and internal validation sets for predictive model development using ML algorithms, with subsequent external validation in both Medical Information Mart for Intensive Care IV (MIMIC-IV) and Medical Information Mart for Intensive Care III (MIMIC-III) databases to identify the optimal model; and (3) SHapley Additive exPlanations (SHAP) value analysis to enhance the clinical interpretability of the selected model 12 . This study rigorously adhered to two international reporting guidelines: the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement for multidimensional confounding adjustment and risk quantification analyses, and the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) statement for model development and reporting 13,14 . 2.2 Date source The data of this study were derived from three major critical care databases: the eICU database (v2.0, containing >200,000 ICU admissions across the US from 2014 to 2015), MIMIC-IV (v3.1, comprising >65,000 ICU and >200,000 ED patients from Beth Israel Deaconess Medical Center with modular design, covering data from 2008 to 2022), and the CareVue subset of MIMIC-III (containing data from 2001 to 2008, excluding temporal overlap with MIMIC-IV) 15-17 .One of the authors of this study has completed the Collaborative Institutional Training Initiative (CITI) Program certification (Certificate ID: 46212703) and obtained official authorization to access these three databases for retrospective research purposes. The eICU data was utilized to demonstrate that comorbid malignancies significantly increased in-hospital mortality among HF patients admitted to the ICU. These data were partitioned into training and internal validation sets for ML model development, while MIMIC-IV and MIMIC-III data served for external validation. 2.3 Participants This study identified eligible participants using International Classification of Diseases, Ninth/Tenth Revision (ICD-9/10) codes. For the multidimensional confounding adjustment and risk quantification analysis, we included all adult (age ≥18 years)HF patients admitted to the ICU, with the following exclusion criteria: (1) ICU length of stay <24 hours; (2) Pregnancy or lactation; and (3) non-first-time ICU admissions. For the ML modeling phase, the study population was restricted to HF patients with malignancy comorbidity while applying the same exclusion criteria to maintain cohort consistency. 2.4 Data extraction and outcomes We extracted comprehensive patient data from the eICU and MIMIC databases, including: (1) demographic characteristics (gender, age, race and weight); (2) illness severity scores (SAPS-II); (3) comorbidities (hypertension, atrial fibrillation, coronary artery disease, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, and stroke); (4) critical interventions within the first 24 hours of ICU admission (mechanical ventilation, continuous renal replacement therapy CRRT, sedative use, and vasopressors administration); (5) vital signs upon ICU admission (mean arterial pressure, heart rate, temperature, respiratory rate, and peripheral oxygen saturation); and (6) laboratory tests (bicarbonate, creatinine, blood urea nitrogen, white blood cell count, red cell distribution width(RDW), hemoglobin, platelet count, calcium, chloride, sodium, potassium, and blood glucose levels).The outcome of this study was in-hospital mortality. 2.5 Data Processing In the data cleaning process, variables with missing values exceeding 30% were removed. For variables with missing values below 30%, we applied k-nearest neighbors (KNN) imputation 18 . Outliers were handled using winsorization (capping): values exceeding the 99th percentile were replaced with the 99th percentile value, and values below the 1st percentile were replaced with the 1st percentile value. Additionally, zero-variance and near-zero-variance variables were excluded. Multicollinearity was assessed to ensure no high correlations existed among the remaining variables. For the external validation phase using the MIMIC-IV and MIMIC-III databases, we adopted a complete-case analysis approach, excluding samples with any missing values to ensure robustness and comparability across datasets. 2.6 Association Analysis Methods This study employed multiple analytical approaches to examine the relationship between malignancies and patient outcomes. Initially, propensity score matching (PSM) 19 was performed to adjust for confounding factors using covariate balancing propensity score (CBPS) estimation 20 . Propensity scores were calculated via CBPS, followed by 1:2 nearest-neighbor matching with a caliper width of 0.1. The effectiveness of PSM was evaluated by calculating standardized mean differences (SMDs), with successful balance defined as all post-matching SMDs <0.1. Successful balance was confirmed when all post-matching standardized mean differences (SMDs) of covariates were <0.1. Weighted logistic regression models were then constructed using the matched sample to estimate effect sizes. To verify result robustness, supplementary analyses were conducted using multivariable logistic regression and IPTW. The multivariable model incorporated all baseline covariates to examine the significance of malignancy-outcome associations. IPTW analysis employed stabilized weights derived from propensity scores, with satisfactory covariate balance achieved after weighting. Consistent findings across PSM, multivariable regression, and IPTW analyses (when regression results from weighted data aligned with both PSM and conventional multivariable results) further reinforced the reliability of our conclusions. This methodological triangulation strategy enhanced the reliability of our observational analyses while maintaining appropriate interpretation boundaries. 2.7 Model development and evaluation This study employed ML approaches to develop predictive models, initially partitioning the eICU database into training and internal validation sets at a 7:3 ratio while addressing class imbalance in the training set using the ROSE algorithm. To prevent overfitting, key variables were selected through the Boruta feature selection method. Subsequently, five ML algorithms—including eXtreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost), k-Nearest Neighbors (KNN), Naïve Bayes (NB), and Logistic Regression (LR)—were implemented, with hyperparameter optimization achieved via 10-fold cross-validation. Model performance was comprehensively evaluated using multiple metrics, including receiver operating characteristic (ROC) curves, Brier score, accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1-score, across both internal validation and two external validation datasets to ensure predictive accuracy and generalizability. Upon completion of model development, we will employ the SHAP method for dual-perspective interpretability analysis: First, global interpretation will be achieved by calculating mean SHAP values across the test cohort to identify the most influential clinical features affecting prediction outcomes; second, individualized SHAP force plots will be generated for both typical cases and special scenarios to visually demonstrate how each feature contributes to prediction results for specific cases. All visualization outputs will be optimized according to clinicians' cognitive characteristics to ensure clinical interpretability and practical utility of the model predictions. This dual-analysis strategy not only validates the biological plausibility of the model at the population level but also provides transparent decision-making references at the individual level 21 . Furthermore, a web-based application was developed to facilitate clinical implementation of the final model. All statistical analyses were conducted using R software (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria), with two-sided *p*-values < 0.05 considered statistically significant. 3. Results Figure 1 illustrates the technical flowchart of this study, with the left panel depicting the association analysis component. Initially, we identified 28,449 HF patients from the eICU database using ICD codes. After applying stringent inclusion and exclusion criteria, 21,636 patients were enrolled in the study cohort, including 18,239 without a history of malignant tumors and 3,397 with comorbid malignancies. Following association analysis, the 3,397 malignancy-complicated HF patients were randomly split into a training set (2,368 patients for ML model development) and an internal validation set at a 7:3 ratio. To further evaluate model performance, external validation cohorts comprising 1,817 and 539 patients were additionally recruited from the MIMIC-IV and MIMIC-III databases, respectively. 3.1 Baseline characteristics The baseline characteristics before PSM are presented in Table 1. Patients with malignancy exhibited significantly higher disease severity on ICU admission, as reflected by SAPS-II scores (median 35 (IQR: 28–44) vs. 32 (24–41)). They were also older (median age 75 (67–83) vs. 71 (61–81)) and had a higher prevalence of atrial fibrillation (45% vs. 39%) and chronic obstructive pulmonary disease (38% vs. 34%). Laboratory tests differences included lower hemoglobin levels (10.2 g/dL (8.9–11.5) vs. 10.7 (9.3–12.1)) and elevated RDW (15.9 (14.7–17.4) vs. 15.5 (14.5–17.0)). No significant differences were observed in mechanical ventilation (46% vs. 50%) or vasopressor use (20% vs. 20%) within the first 24 hours of ICU admission. 3.2 Association Analysis Results Here are the results of our association analysis using PSM and weighted logistic regression. We performed 1:2 PSM with a caliper set to ensure SMD <0.1 for all covariates, achieving good balance between the malignancy and non-malignancy groups in the matched cohort (Table 1, Figure2-A). The weighted logistic regression model, adjusting for all baseline characteristics including demographics, severity scores, comorbidities, vital signs, and laboratory values, demonstrated that malignancy was significantly associated with worse hospital outcomes (OR 1.14, 95% CI 1.02-1.26, p=0.0177). This analysis accounted for the matched nature of the data through robust variance estimation using subclass clustering, and the quasibinomial family was used to address potential overdispersion. The results suggest that malignancy independently predicts poorer outcomes in this critically ill population even after rigorous adjustment for potential confounders through both matching and regression techniques. To further validate our findings, we conducted sensitivity analyses using IPTW and multivariate logistic regression. The IPTW approach achieved excellent covariate balance (all SMD <0.1) as demonstrated in the Love plot (Figure2-B), and the weighted analysis similarly showed that malignancy was associated with increased mortality risk (OR 1.16, 95% CI 1.03-1.30, p=0.0132). The traditional multivariate logistic regression model, adjusting for all baseline covariates, yielded consistent results (OR 1.20, 95% CI 1.07-1.35, p=0.00224). Figure 3 presents the forest plot comparing these three association analysis methods (PSM, IPTW, and multivariate logistic regression), demonstrating robust and consistent effect estimates across different analytical approaches. The concordance of results from these complementary methods strengthens the evidence for an independent association between malignancy and poorer outcomes in critically ill patients. 3.3 Machine Learning Modeling – Feature Selection Using the Boruta algorithm for all-cause mortality prediction in HF patients with malignancy, we identified 19 key features from comprehensive clinical variables（Figure 4）. The confirmed significant features included illness severity scores (SAPS-II), critical interventions (mechanical ventilation, sedatives, vasopressors), comorbidities (atrial fibrillation), vital signs (mean arterial pressure, heart rate, temperature, respiratory rate, peripheral oxygen saturation), and laboratory tests (bicarbonate, creatinine, blood urea nitrogen, white blood cell count, hemoglobin, platelet count, and electrolytes including chloride, sodium, and potassium). 3.4 Model Construction and Comparative Performance Evaluation During the model construction phase, we employed five ML algorithmsto develop predictive models. The eICU dataset was partitioned into training and internal validation sets at a 7:3 ratio. To address class imbalance in the training set, we applied the Random Over-Sampling Examples (ROSE) algorithm for data resampling. Model training was optimized through a 10-fold cross-validation approach combined with grid search to systematically tune hyperparameters, ensuring robust performance and generalizability. We comprehensively evaluated model performance using multiple metrics, conducting internal validation with the partitioned eICU dataset and external validation with both MIMIC-IV and MIMIC-III databases. As illustrated in Figure 5, the ROC curves demonstrate distinct performance patterns across the training set (A), internal validation set (B), MIMIC-IV external validation set (C), and MIMIC-III external validation set (D). While XGBoost achieved superior AUC in the training set, its performance significantly declined across all validation sets, indicating evident overfitting. In contrast, AdaBoost exhibited the most robust and consistent performance across all datasets, particularly maintaining superior predictive accuracy during external validation, suggesting better generalizability for clinical application. As demonstrated in Table 2, the AdaBoost model demonstrated superior performance in predicting mortality among HF patients across multiple datasets. In external validation (MIMIC-III/IV), AdaBoost achieved the best balance between calibration (Brier Score: 0.187–0.214) and discrimination (Accuracy: 0.672–0.720), outperforming LR and KNN. It exhibited high sensitivity (0.745–0.814) to minimize false negatives and the highest PPV (0.829–0.902) for reliable positive predictions. While XGBoost showed overfitting (F1 drop from 0.983 to 0.853 in validation), AdaBoost maintained robust generalization (F1: 0.810–0.856). Its stability across datasets supports clinical deployment for mortality risk stratification. 3.5 SHAP -Based Interpretability Assessment To enhance the model's interpretability, we performed SHAP analysis on the AdaBoost model. Figure 6 presents the global interpretation plots from SHAP analysis, where Figure 6-A displays the beeswarm plot with model features on the y-axis and their corresponding SHAP values on the x-axis. In the beeswarm plot, yellow dots indicate strong positive contributions to predictions while purple dots represent negative or minimal impacts, with wider horizontal distributions reflecting greater variability in feature effects. Figure 6-B shows the feature importance plot, which ranks variables by their mean absolute SHAP values, providing a comprehensive visualization of each feature's global impact on model outputs. As illustrated in Figure 6, the top five most influential predictive variables were identified as the SAPS-II score (Simplified Acute Physiology Score II), requirement for mechanical ventilation, heart rate, body temperature, and respiratory rate, which collectively demonstrated the strongest associations with clinical outcomes in our predictive model. Figure 7 presents the local interpretability analysis using SHAP force plots, illustrating the model's prediction process for an individual patient. The final prediction output (f(x)=1) demonstrates that negative cumulative effects lowered the predicted value below the baseline, classifying this patient as a survivor. SHAP value analysis identified SAPS-II score, ventilation status, and SpO2 levels as the primary determinants driving this prediction, with each factor's relative contribution quantitatively visualized in the force plot representation. 3.6 Interactive Model Visualization for Clinical Decision Support To facilitate clinical application, we developed a user-friendly web-based calculator (available at: https://nanzihan1998.shinyapps.io/Mortality/) that enables real-time mortality risk assessment for HF patients with comorbid malignancy. 4. Discussion This study comprises two integrated components: association analysis and ML. The first component systematically employs three distinct association analysis methodologies to conclusively demonstrate that HF patients with comorbid malignancy experience significantly worse clinical outcomes. Three complementary association analysis approaches consistently demonstrated that malignancy significantly increased in-hospital mortality among ICU-admitted HF patients: PSM (OR 1.14, 95% CI [1.02–1.26]), IPTW (OR 1.16, 95% CI [1.03–1.30]), and multivariate logistic regression (OR 1.20, 95% CI [1.07–1.35]). The concordant results across all methods robustly confirm the detrimental impact of malignancy on clinical outcomes in this vulnerable population. The poorer prognosis observed in patients with comorbid malignancies stems from multiple factors, with treatment-related cardiotoxicity representing one of the significant contributors. Anthracyclines, cornerstone agents for various solid and hematologic malignancies 22 , induce dose-dependent cardiomyocyte damage through topoisomerase 2β inhibition, activating cell death pathways and impairing mitochondrial biogenesis 23 , 24 . Other chemotherapeutic agents including 5-fluorouracil 25 , cyclophosphamide 26 , and paclitaxel 27 similarly demonstrate cardiotoxic effects. Beyond chemotherapy, immune checkpoint inhibitors (ICIs) represent a breakthrough in cancer therapy by blocking PD-1/PD-L1 or CTLA-4 to restore antitumor immunity 28 . However, ICIs frequently induce inflammatory toxicities termed immune-related adverse events (irAEs) 29 . Unlike chemotherapy-induced organ damage, irAEs exhibit unpredictable onset and severity 29 . Cardiac involvement occurs in 1.14% of ICI recipients, increasing to 2.4% with combination therapy 30 . ICI-associated cardiotoxicity correlates with elevated mortality and often necessitates treatment interruption, requiring high-dose corticosteroids for management 31 . In addition to treatment-related cardiotoxicity, cardiac metastasis - particularly to the myocardium or pericardium - can directly compromise cardiac structure and function, exacerbating HF severity. Furthermore, the metabolic demands of rapidly proliferating tumor cells induce a catabolic state characterized by hypoalbuminemia and anemia. This tumor-associated malnutrition exacerbates myocardial energy deficiency in HF patients, whose compromised cardiac output already limits nutrient delivery. The resultant impairment of cardiac repair mechanisms creates a vicious cycle that accelerates disease progression. The second component focuses specifically on HF patients with comorbid malignancy, establishing a ML-based predictive model for in-hospital mortality among this high-risk population admitted to ICU. The predictive model was developed using five ML algorithms incorporating 19 carefully selected clinical variables from the first 24 hours of ICU admission. Comparative evaluation demonstrated AdaBoost's superior performance, with Youden's index identifying patients having a predicted probability > 0.483 as high-risk for in-hospital mortality. To enhance model credibility, SHAP analysis was employed for comprehensive interpretation - the summary plot provided global visualization of all 19 risk factors' contributions, while force plots enabled individualized prediction explanations for clinical implementation. The SAPS-II score emerged as the most significant predictive variable in our model 32 . As a well-validated scoring system specifically designed to assess mortality risk in critically ill patients, SAPS-II provide a comprehensive evaluation of illness severity and prognostic outcomes. Its widespread adoption in ICUs and relative ease of acquisition further enhance its clinical utility for risk stratification. Mechanical ventilation requirement ranked as the second most important predictor, with the beeswarm plot demonstrating significantly worse outcomes in ventilated patients. In HF, elevated left ventricular filling pressures frequently lead to alveolar pulmonary edema, impairing oxygenation and ventilation - a pathophysiological cascade culminating in respiratory failure 33 . Substantial evidence indicates that HF patients complicated by respiratory failure exhibit poorer prognoses, warranting prompt positive-pressure mechanical ventilation to improve clinical outcomes 34 . Our study further identified elevated heart rate as an independent predictor of poorer outcomes, a finding consistent with prior evidence. For patients with tachycardia, pharmacological interventions including β-blockers and ivabradine may be considered to attenuate sympathetic overactivation, thereby potentially improving clinical prognosis 35 – 37 . Notably, observations in heart transplant recipients have also demonstrated a significant correlation between elevated heart rate and deterioration of cardiac function, reinforcing the pathophysiological relevance of heart rate modulation in cardiac patients 38 . Hypothermia has been consistently associated with adverse clinical outcomes in HF patients. This association may stem from severely compromised cardiac output in advanced disease stages, leading to systemic hypoperfusion, impaired tissue oxygenation, and consequent thermoregulatory failure - a maladaptive cascade that exacerbates end-organ dysfunction. Our model additionally incorporated clinically relevant variables from the following categories: comorbidities (atrial fibrillation); critical interventions within the first 24 hours of ICU admission (vasoactive drugs and sedatives); vital signs (respiratory rate, peripheral oxygen saturation, mean arterial pressure); and laboratory tests (white blood cell count, blood urea nitrogen, creatinine, hemoglobin, bicarbonate, chloride, sodium, potassium, and platelet count). This study possesses several notable merits. First, it represents the first investigation to systematically demonstrate worse prognosis in HF patients with comorbid malignancy using three distinct association analysis methodologies. Second, we developed a ML-based predictive model for in-hospital mortality among ICU-admitted HF patients with malignancy, incorporating SHAP analysis to enhance model interpretability and facilitate clinical translation of artificial intelligence technology. Third, the visualization of our model enables user-friendly clinical application, supporting early mortality risk assessment to inform clinical decision-making. Several limitations should be acknowledged. As a multicenter retrospective study, inherent issues of missing data and outliers exist, though we employed rigorous statistical methods to maximize data validity. Our model currently utilizes only cross-sectional data from the first ICU day; future incorporation of longitudinal data may improve predictive accuracy. Additionally, the exclusion of novel biomarkers such as Growth Differentiation Factor-15, mid-regional proatrial natriuretic peptide, and ceruloplasmin may limit predictive performance, suggesting an avenue for model refinement 39 . 5. Conclusions Through robust association analysis, we demonstrated that malignancy significantly worsens outcomes in HF patients. We developed and validated a ML model to predict in-hospital mortality risk in this high-risk population. SHAP analysis enhanced model interpretability, enabling clinicians to identify key mortality predictors and facilitate timely interventions to improve patient prognosis. Abbreviations AdaBoost: Adaptive Boosting; CBPS: covariate balancing propensity score; CRRT: continuous renal replacement therapy; eICU: eICU Collaborative Research Database; HF: heart failure; ICD: International Classification of Diseases; ICI: immune checkpoint inhibitor; ICU: Intensive Care Unit; IPTW: inverse probability treatment weighting; KNN: k-nearest neighbors; LR: Logistic Regression; MIMIC: Medical Information Mart for Intensive Care; ML: machine learning; NB: Naïve Bayes; PSM: propensity score matching; RDW: red cell distribution width; ROSE: Random Over-Sampling Examples; SAPS-II: Simplified Acute Physiology Score II; SHAP: SHapley Additive exPlanations; SMD: standardized mean difference; SpO2: peripheral oxygen saturation; STROBE: STrengthening the Reporting of OBservational studies in Epidemiology; TRIPOD: Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis; XGBoost: eXtreme Gradient Boosting. Declarations Acknowledgments We would like to express our sincere gratitude to the official operators of the MIMIC and eICU databases for providing access to these valuable clinical datasets. This study was supported by the Health Commission of Hebei Province and the Department of Science and Technology of Hebei Province. Their support was instrumental in facilitating this research endeavor. Authors’ contributions Tao Zhang and Zihan Nan created the study protocol and wrote the first manuscript draft. Zihan Nan performed all data collection and statistical analyses. Shaohan Guo conceived the study and critically revised the manuscript. Shiju Yang assisted with the study design. Leying Li assisted with manuscript editing. Yifei Zhang assisted with the interpretation of statistical methods. Yan Xin assisted with manuscript revision and data validation. Zhenjie Hu and Congcong Zhao supervised the study, contributed to data interpretation, and finalized the manuscript. All authors read and approved the final manuscript. Funding This work was supported by the Medical Science Research Project of Hebei (Grant No. 20250673) and the S&T Program of Hebei (Grant No. 223777104D). Availability of data and materials The data extracted in this study are available from the corresponding authors upon reasonable request. Ethics approval and consent to participate The eICU database and MIMIC-III/IV databases were also approved by the Massachusetts Institute of Technology (Cambridge, MA) and Beth Israel Deaconess Medical Center (Boston, MA), and consent was obtained for the original data collection. Consent for publication Not applicable Competing interests The authors declare no conflicts of interest. References Koene, R. J., Prizment, A. E., Blaes, A. & Konety, S. H. Shared Risk Factors in Cardiovascular Disease and Cancer. Circulation 133 , 1104-1114, doi:10.1161/CIRCULATIONAHA.115.020406 (2016). Khan, M. S. et al. Global epidemiology of heart failure. Nat Rev Cardiol 21 , 717-734, doi:10.1038/s41569-024-01046-6 (2024). Wang, H. et al. Mortality in patients admitted to hospital with heart failure in China: a nationwide Cardiovascular Association Database-Heart Failure Centre Registry cohort study. Lancet Glob Health 12 , e611-e622, doi:10.1016/S2214-109X(23)00605-8 (2024). Wu, Z., Xia, F. & Lin, R. Global burden of cancer and associated risk factors in 204 countries and territories, 1980-2021: a systematic analysis for the GBD 2021. J Hematol Oncol 17 , 119, doi:10.1186/s13045-024-01640-8 (2024). Nohria, A. Double Jeopardy: Cancer and Heart Failure. J Card Fail 25 , 522-523, doi:10.1016/j.cardfail.2019.05.002 (2019). Tuzovic, M. et al. National Outcomes in Hospitalized Patients With Cancer and Comorbid Heart Failure. J Card Fail 25 , 516-521, doi:10.1016/j.cardfail.2019.02.007 (2019). Li, J. et al. Predicting Mortality in Intensive Care Unit Patients With Heart Failure Using an Interpretable Machine Learning Model: Retrospective Cohort Study. J Med Internet Res 24 , e38082, doi:10.2196/38082 (2022). Peng, S. et al. Interpretable machine learning for 28-day all-cause in-hospital mortality prediction in critically ill patients with heart failure combined with hypertension: A retrospective cohort study based on medical information mart for intensive care database-IV and eICU databases. Front Cardiovasc Med 9 , 994359, doi:10.3389/fcvm.2022.994359 (2022). Vinter, N. et al. Development and validation of prediction models for incident atrial fibrillation in heart failure. Open Heart 10 , doi:10.1136/openhrt-2022-002169 (2023). Hounye, A. H., Xiong, L. & Hou, M. Integrated explainable machine learning and multi-omics analysis for survival prediction in cancer with immunotherapy response. Apoptosis 30 , 364-388, doi:10.1007/s10495-024-02050-4 (2024). Montazeri, M., Montazeri, M., Montazeri, M. & Beigzadeh, A. Machine learning models in breast cancer survival prediction. Technology and Health Care 24 , 31-42, doi:10.3233/thc-151071 (2016). Lundberg, S. M. & Lee, S.-I. A unified approach to interpreting model predictions. Advances in neural information processing systems 30 (2017). Vandenbroucke, J. P. et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med 4 , e297, doi:10.1371/journal.pmed.0040297 (2007). Collins, G. S., Reitsma, J. B., Altman, D. G. & Moons, K. G. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ 350 , g7594, doi:10.1136/bmj.g7594 (2015). Pollard, T. J. et al. The eICU Collaborative Research Database, a freely available multi-center database for critical care research. Sci Data 5 , 180178, doi:10.1038/sdata.2018.178 (2018). Johnson, A. E. W. et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data 10 , 1, doi:10.1038/s41597-022-01899-x (2023). Johnson, A. E. et al. MIMIC-III, a freely accessible critical care database. Sci Data 3 , 160035, doi:10.1038/sdata.2016.35 (2016). Pujianto, U., Wibawa, A. P. & Akbar, M. I. in 2019 5th International Conference on Science in Information Technology (ICSITech). 83-88 (IEEE). Haukoos, J. S. & Lewis, R. J. The propensity score. Jama 314 , 1637-1638 (2015). Imai, K. & Ratkovic, M. Covariate balancing propensity score. Journal of the Royal Statistical Society Series B: Statistical Methodology 76 , 243-263 (2014). Lundberg, S. M. et al. From local explanations to global understanding with explainable AI for trees. Nature Machine Intelligence 2 , 56-67, doi:10.1038/s42256-019-0138-9 (2020). Bloom, M. W. et al. Cardio-Oncology and Heart Failure: a Scientific Statement From the Heart Failure Society of America. J Card Fail 31 , 415-455, doi:10.1016/j.cardfail.2024.08.045 (2025). Henriksen, P. A. Anthracycline cardiotoxicity: an update on mechanisms, monitoring and prevention. Heart 104 , 971-977, doi:10.1136/heartjnl-2017-312103 (2018). Agunbiade, T. A., Zaghlol, R. Y. & Barac, A. Heart Failure in Relation to Anthracyclines and Other Chemotherapies. Methodist Debakey Cardiovasc J 15 , 243-249, doi:10.14797/mdcj-15-4-243 (2019). Gradishar, W. J. & Vokes, E. E. 5-Fluorouracil cardiotoxicity: a critical review. Ann Oncol 1 , 409-414, doi:10.1093/oxfordjournals.annonc.a057793 (1990). Iqubal, A. et al. Molecular mechanism involved in cyclophosphamide-induced cardiotoxicity: Old drug with a new vision. Life Sci 218 , 112-131, doi:10.1016/j.lfs.2018.12.018 (2019). Perez, E. A. Paclitaxel and cardiotoxicity. J Clin Oncol 16 , 3481-3482, doi:10.1200/JCO.1998.16.11.3481 (1998). Arafat Hossain, M. A comprehensive review of immune checkpoint inhibitors for cancer treatment. Int Immunopharmacol 143 , 113365, doi:10.1016/j.intimp.2024.113365 (2024). Wang, S. J., Dougan, S. K. & Dougan, M. Immune mechanisms of toxicity from checkpoint inhibitors. Trends Cancer 9 , 543-553, doi:10.1016/j.trecan.2023.04.002 (2023). Mahmood, S. S. et al. Myocarditis in Patients Treated With Immune Checkpoint Inhibitors. J Am Coll Cardiol 71 , 1755-1764, doi:10.1016/j.jacc.2018.02.037 (2018). Patel, R. P. et al. Cardiotoxicity of Immune Checkpoint Inhibitors. Curr Oncol Rep 23 , 79, doi:10.1007/s11912-021-01070-6 (2021). Keegan, M. T., Gajic, O. & Afessa, B. Severity of illness scoring systems in the intensive care unit. Crit Care Med 39 , 163-169, doi:10.1097/CCM.0b013e3181f96f81 (2011). Alviar, C. L. et al. Positive Pressure Ventilation in Cardiogenic Shock: Review of the Evidence and Practical Advice for Patients With Mechanical Circulatory Support. Can J Cardiol 36 , 300-312, doi:10.1016/j.cjca.2019.11.038 (2020). Tavazzi, G. Mechanical ventilation in cardiogenic shock. Curr Opin Crit Care 27 , 447-453, doi:10.1097/MCC.0000000000000836 (2021). Kapoor, J. R. & Heidenreich, P. A. Role of heart rate as a marker and mediator of poor outcome for patients with heart failure. Curr Heart Fail Rep 9 , 133-138, doi:10.1007/s11897-012-0086-8 (2012). Sciatti, E., Vizzardi, E., Bonadei, I., Dallapellegrina, L. & Carubelli, V. The role of heart rate and ivabradine in acute heart failure. Monaldi Arch Chest Dis 89 , doi:10.4081/monaldi.2019.1091 (2019). Kubrychtova, V. et al. Heart rate recovery and prognosis in heart failure patients. Eur J Appl Physiol 105 , 37-45, doi:10.1007/s00421-008-0870-z (2009). Liebo, M. et al. Elevated Heart Rate Following Heart Transplantation Is Associated With Increased Graft Vasculopathy and Mortality. J Card Fail 25 , 249-256, doi:10.1016/j.cardfail.2019.01.009 (2019). Savic-Radojevic, A. et al. Novel Biomarkers of Heart Failure. Adv Clin Chem 79 , 93-152, doi:10.1016/bs.acc.2016.09.002 (2017). Tables Tables 1 and 2 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Table1.docx Table2.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 04 Nov, 2025 Reviews received at journal 24 Sep, 2025 Reviewers agreed at journal 10 Sep, 2025 Reviewers agreed at journal 09 Sep, 2025 Reviews received at journal 31 Aug, 2025 Reviewers agreed at journal 16 Aug, 2025 Reviewers agreed at journal 13 Aug, 2025 Reviewers invited by journal 13 Aug, 2025 Editor assigned by journal 13 Aug, 2025 Editor invited by journal 28 Jul, 2025 Submission checks completed at journal 24 Jul, 2025 First submitted to journal 24 Jul, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7179137","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":501739506,"identity":"197137ec-956d-409c-b12f-f021450f9132","order_by":0,"name":"Tao Zhang","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Tao","middleName":"","lastName":"Zhang","suffix":""},{"id":501739507,"identity":"d3e25749-1772-4f5e-a801-fe7350c669a1","order_by":1,"name":"Zihan Nan","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zihan","middleName":"","lastName":"Nan","suffix":""},{"id":501739508,"identity":"8de70784-6b90-46a3-bd00-6d906a7bc351","order_by":2,"name":"Shaohan Guo","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shaohan","middleName":"","lastName":"Guo","suffix":""},{"id":501739513,"identity":"66e9a2f0-487a-4d49-aa54-1ea4d94f6040","order_by":3,"name":"Shiju Yang","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shiju","middleName":"","lastName":"Yang","suffix":""},{"id":501739514,"identity":"ee84c806-7e69-46ca-b115-4e73789eb694","order_by":4,"name":"Leying Li","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Leying","middleName":"","lastName":"Li","suffix":""},{"id":501739515,"identity":"3d3e3406-46e5-459f-8c27-b784e0610bec","order_by":5,"name":"Yifei Zhang","email":"","orcid":"","institution":"Hebei General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yifei","middleName":"","lastName":"Zhang","suffix":""},{"id":501739517,"identity":"cb629df8-5491-4a35-b401-8cd0d013ef61","order_by":6,"name":"Yan Xin","email":"","orcid":"","institution":"Shijiazhuang Third Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Xin","suffix":""},{"id":501739518,"identity":"cb0ca637-aaa5-4224-bc2f-d2b08962ecc5","order_by":7,"name":"Zhenjie Hu","email":"","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhenjie","middleName":"","lastName":"Hu","suffix":""},{"id":501739519,"identity":"878711bd-718a-4b70-a05b-ffeed1afa289","order_by":8,"name":"Congcong Zhao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIiWNgGAWjYFACxgYwxS/BwAYROECsFskZxGuBAoMbxGqRn5Hc+Jnnj5288e3mY49utjHI8d1IYPxcgNfwxGZpHp5kw213jqUb57YxGEveSGCWnoFPi0RiGzOPxIEEsxs5ZtJALYkbbiSwMfPgdRhIi8GBBOMZ+d9AWuoJamG4AdKScCDBQCKHDaQlwYCQFoMzD5sl5xxINpxxI83cOOechOFMoIg0Xoe1pz/88AYYYvwzkp89zimzkec7nnzwM16HCSSgcCUY4JGLE/AfwC8/CkbBKBgFo4ABALpZSbW/z7BJAAAAAElFTkSuQmCC","orcid":"","institution":"The Fourth Hospital of Hebei Medical University","correspondingAuthor":true,"prefix":"","firstName":"Congcong","middleName":"","lastName":"Zhao","suffix":""}],"badges":[],"createdAt":"2025-07-21 15:38:36","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7179137/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7179137/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89564331,"identity":"4f457e13-3412-4851-968d-5d7c6a21ca26","added_by":"auto","created_at":"2025-08-21 10:33:42","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":77610,"visible":true,"origin":"","legend":"\u003cp\u003eTechnical flowchart.\u003c/p\u003e\n\u003cp\u003eLegend: eICU: eICU Collaborative Research Database; ICU: Intensive Care Unit; MIMIC-III: Medical Information Mart for Intensive Care III; MIMIC-IV: Medical Information Mart for Intensive Care IV\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/2a03d8164feb42916e317e0e.jpg"},{"id":89562521,"identity":"2083d5cb-059c-40b0-9b0e-fbfcaef5228b","added_by":"auto","created_at":"2025-08-21 10:25:42","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":79704,"visible":true,"origin":"","legend":"\u003cp\u003eAssessment of covariate balance between malignancy and non-malignancy groups after propensity score matching (A) and inverse probability treatment weighting (B).\u003c/p\u003e\n\u003cp\u003eLegend: AFib: Atrial Fibrillation; BUN: Blood Urea Nitrogen; CAD: Coronary Artery Disease; CKD: Chronic Kidney Disease; COPD: Chronic Obstructive Pulmonary Disease; CRRT: Continuous Renal Replacement Therapy; Hb: Hemoglobin; HCO3: Bicarbonate; HR: Heart Rate; MAP: Mean Arterial Pressure; PSM: Propensity Score Matching; IPTW: Inverse Probability Treatment Weighting; RDW: Red Cell Distribution Width; RR: Respiratory Rate; SAPS-II: Simplified Acute Physiology Score II; SpO2: Oxygen Saturation; WBC: White Blood Cell Count\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/ab365b3f1b2efa6a1a63e35e.jpg"},{"id":89562520,"identity":"9835f3cc-3870-41ab-9dd2-e8929f1e620a","added_by":"auto","created_at":"2025-08-21 10:25:42","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":40823,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of malignancy-associated mortality risk estimates across different adjustment methods.\u003c/p\u003e\n\u003cp\u003eLegend: PSM: Propensity Score Matching; IPTW: Inverse Probability Treatment Weighting; OR (95% CI): Odds Ratio (95% Confidence Interval)\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/2946205b4af3e9868e9d9611.jpg"},{"id":89564329,"identity":"0b95e67e-f5f1-43c3-a0fd-ff80bb32bfed","added_by":"auto","created_at":"2025-08-21 10:33:42","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":86550,"visible":true,"origin":"","legend":"\u003cp\u003eBoruta-Selected features for mortality prediction in heart failure patients with malignancy.\u003c/p\u003e\n\u003cp\u003eLegend: AFib: Atrial Fibrillation; BUN: Blood Urea Nitrogen; CAD: Coronary Artery Disease; CKD: Chronic Kidney Disease; COPD: Chronic Obstructive Pulmonary Disease; CRRT: Continuous Renal Replacement Therapy; Hb: Hemoglobin; HCO3: Bicarbonate; HR: Heart Rate; MAP: Mean Arterial Pressure; RDW: Red Cell Distribution Width; RR: Respiratory Rate; SAPS-II: Simplified Acute Physiology Score II; SpO2: Oxygen Saturation; WBC: White Blood Cell Count\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/e18d418b72df29b2801b0843.jpg"},{"id":89562527,"identity":"cb724e51-cff7-4f38-896c-e69464bae8b7","added_by":"auto","created_at":"2025-08-21 10:25:42","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":82617,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of machine learning models in (A) training set, (B) internal validation set, (C) MIMIC-IV external validation, and (D) MIMIC-III external validation\u003c/p\u003e\n\u003cp\u003eLegend: AUC: Area Under the Curve; XGBoost: Extreme Gradient Boosting; AdaBoost: Adaptive Boosting; KNN: k-Nearest Neighbors; LR: Logistic Regression; ROC: Receiver Operating Characteristic\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/bcfbe09da9c68c215b570825.jpg"},{"id":89565643,"identity":"8ccff2ac-bf6a-4348-89a2-bb4e85d5dff6","added_by":"auto","created_at":"2025-08-21 10:49:42","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":63945,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP-based interpretability analysis of the AdaBoost model: (A) Beeswarm plot of feature contributions and (B) Mean absolute SHAP values for feature importance\u003c/p\u003e\n\u003cp\u003eLegend: AFib: Atrial Fibrillation; BUN: Blood Urea Nitrogen; HCO3: Bicarbonate; Hb: Hemoglobin; HR: Heart Rate; MAP: Mean Arterial Pressure; RR: Respiratory Rate; SAPS-II: Simplified Acute Physiology Score II; SHAP: SHapley Additive exPlanations; SpO2: Oxygen Saturation; WBC: White Blood Cell count.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/dfd3df0b8e6fd66a44888630.jpg"},{"id":89562532,"identity":"ec1218f2-418f-4837-8b23-0e681782f1ee","added_by":"auto","created_at":"2025-08-21 10:25:42","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":34658,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP force plot analysis of individual mortality risk prediction\u003c/p\u003e\n\u003cp\u003eLegend: RR: Respiratory Rate; SAPS-II: Simplified Acute Physiology Score II; SpO2: Oxygen Saturation\u003c/p\u003e","description":"","filename":"7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/a2957aa31f1d12bc63487d17.jpg"},{"id":89565645,"identity":"9d38b557-a0ca-480b-97f0-0c853bcf5f9c","added_by":"auto","created_at":"2025-08-21 10:49:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1082597,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/9e90ef48-f169-4325-a6dc-7215b1b38718.pdf"},{"id":89565368,"identity":"c33c83a2-2b7c-47ff-aee5-20b5b73c4ca2","added_by":"auto","created_at":"2025-08-21 10:41:42","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":23453,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/5abe71c226324afa391d7cc4.docx"},{"id":89562523,"identity":"7bbffa07-2206-4533-855c-ee82a2ab79d9","added_by":"auto","created_at":"2025-08-21 10:25:42","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":20411,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-7179137/v1/33559d01e3dffb62cc440ada.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multicenter Machine Learning Model for Assessing the Impact of Malignancy on In-Hospital Mortality in Heart Failure Patients: A Clinical Decision Support System with Interpretable Artificial Intelligence","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eHeart failure (HF) and malignancy represent two of the most significant global health challenges, with both conditions showing rising prevalence due to aging populations\u003csup\u003e1\u003c/sup\u003e. HF affects over 56\u0026nbsp;million people worldwide and is associated with high morbidity and mortality, including a 28.2% all-cause mortality rate at 3 years post-diagnosis\u003csup\u003e2,3\u003c/sup\u003e. Meanwhile, malignancy remains the second leading cause of death globally, accounting for 14.57% of total mortality\u003csup\u003e4\u003c/sup\u003e. Notably, these conditions frequently coexist, particularly in older adults\u0026mdash;approximately 10% of patients with malignancy aged\u0026thinsp;\u0026ge;\u0026thinsp;65 years also have HF \u003csup\u003e5\u003c/sup\u003e. This overlap poses a substantial clinical burden, as patients with both HF and malignancy experience worse outcomes, including prolonged hospital stays, increased healthcare costs, and nearly threefold higher in-hospital mortality compared to those with either condition alone\u003csup\u003e6\u003c/sup\u003e. Accurate prognostic prediction in patients with concurrent HF and malignancy is crucial for optimizing therapeutic decision-making, yet remains a significant clinical challenge. The intricate interplay between these conditions\u0026mdash;encompassing shared risk factors, overlapping pathophysiological pathways, and malignancy treatment-related cardiotoxicity\u0026mdash;substantially complicates risk stratification and clinical management. A comprehensive elucidation of malignancy-associated outcomes in HF populations could facilitate more personalized and effective care delivery.\u003c/p\u003e\n\u003cp\u003eRecent advances in data science and artificial intelligence have facilitated the increasing application of machine learning (ML) in prognostic analysis for critically ill Intensive Care Unit (ICU) patients. Especially with the development of interpretable ML techniques, the \u0026quot;black box\u0026quot; of traditional ML has been unveiled, making the model\u0026apos;s results more convincing. In the field of HF, interpretable ML techniques have been widely adopted. For instance, Jili Li\u0026apos;s team employed an interpretable ML model to predict mortality risk in HF patients admitted to the ICU\u003csup\u003e7\u003c/sup\u003e. Similarly, Shengxian Peng et al. developed an interpretable ML approach to forecast 28-day all-cause in-hospital mortality among critically ill HF patients with comorbid hypertension\u003csup\u003e8\u003c/sup\u003e. Additionally, Nicklas Vinter\u0026apos;s team created an interpretable ML model to predict the risk of new-onset atrial fibrillation in HF patients\u003csup\u003e9\u003c/sup\u003e. In the field of oncology, interpretable ML techniques have also demonstrated significant advancements. For example, the NKECLR model developed by Alphonse Houssou Hounye\u0026apos;s team enables precise prediction of survival rates and treatment responses in melanoma patients\u003csup\u003e10\u003c/sup\u003e, while Mitra Montazeri\u0026apos;s research group systematically analyzed survival outcomes across different breast cancer subtypes using ML algorithms\u003csup\u003e11\u003c/sup\u003e. However, in the emerging interdisciplinary field of cardio-oncology, prognostic prediction research specifically targeting HF patients with comorbid malignancies remains notably underdeveloped.\u003c/p\u003e\n\u003cp\u003eTherefore, this study aims to: (1) systematically evaluated the malignancy-associated in-hospital mortality risk in HF patients; (2) develop a prognostic model for predicting in-hospital mortality specifically in HF patients with comorbid malignancy; and (3) conduct interpretability analysis of the model with visualization to facilitate clinical implementation.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cp\u003e2.1 Overview of the Methods\u003c/p\u003e\n\u003cp\u003eThis study employed a three-stage analytical approach: (1) multidimensional confounding adjustment and risk quantification using propensity score matching (PSM), inverse probability treatment weighting (IPTW), and multivariable logistic regression, confirming significantly worse clinical outcomes in HF patients with malignancies; (2) partitioning of the eICU Collaborative Research Database(eICU database) into training and internal validation sets for predictive model development using ML algorithms, with subsequent external validation in both Medical Information Mart for Intensive Care IV (MIMIC-IV) and Medical Information Mart for Intensive Care III (MIMIC-III) databases to identify the optimal model; and (3) SHapley Additive exPlanations (SHAP) value analysis to enhance the clinical interpretability of the selected model\u003csup\u003e12\u003c/sup\u003e. This study rigorously adhered to two international reporting guidelines: the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement for multidimensional confounding adjustment and risk quantification analyses, and the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) statement for model development and reporting\u003csup\u003e13,14\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e2.2 Date source\u003c/p\u003e\n\u003cp\u003eThe data of this study were derived from three major critical care databases: the eICU database (v2.0, containing \u0026gt;200,000 ICU admissions across the US from 2014 to 2015), MIMIC-IV (v3.1, comprising \u0026gt;65,000 ICU and \u0026gt;200,000 ED patients from Beth Israel Deaconess Medical Center with modular design, covering data from 2008 to 2022), and the CareVue subset of MIMIC-III (containing data from 2001 to 2008, excluding temporal overlap with MIMIC-IV) \u003csup\u003e15-17\u003c/sup\u003e.One of the authors of this study has completed the Collaborative Institutional Training Initiative (CITI) Program certification (Certificate ID: 46212703) and obtained official authorization to access these three databases for retrospective research purposes. The eICU data was utilized to demonstrate that comorbid malignancies significantly increased in-hospital mortality among HF patients admitted to the ICU. These data were partitioned into training and internal validation sets for ML model development, while MIMIC-IV and MIMIC-III data served for external validation.\u003c/p\u003e\n\u003cp\u003e2.3 Participants\u003c/p\u003e\n\u003cp\u003eThis study identified eligible participants using International Classification of Diseases, Ninth/Tenth Revision (ICD-9/10) codes. For the multidimensional confounding adjustment and risk quantification analysis, we included all adult (age ≥18 years)HF patients admitted to the ICU, with the following exclusion criteria: (1) ICU length of stay \u0026lt;24 hours; (2) Pregnancy or lactation; and (3) non-first-time ICU admissions. For the ML modeling phase, the study population was restricted to HF patients with malignancy comorbidity while applying the same exclusion criteria to maintain cohort consistency.\u003c/p\u003e\n\u003cp\u003e2.4 Data extraction and outcomes\u003c/p\u003e\n\u003cp\u003eWe extracted comprehensive patient data from the eICU and MIMIC databases, including: (1) demographic characteristics (gender, age, race and weight); (2) illness severity scores (SAPS-II); (3) comorbidities (hypertension, atrial fibrillation, coronary artery disease, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, and stroke); (4) critical interventions within the first 24 hours of ICU admission (mechanical ventilation, continuous renal replacement therapy CRRT, sedative use, and vasopressors administration); (5) vital signs upon ICU admission (mean arterial pressure, heart rate, temperature, respiratory rate, and peripheral oxygen saturation); and (6) laboratory tests (bicarbonate, creatinine, blood urea nitrogen, white blood cell count, red cell distribution width(RDW), hemoglobin, platelet count, calcium, chloride, sodium, potassium, and blood glucose levels).The outcome of this study was in-hospital mortality.\u003c/p\u003e\n\u003cp\u003e2.5 Data Processing\u003c/p\u003e\n\u003cp\u003eIn the data cleaning process, variables with missing values exceeding 30% were removed. For variables with missing values below 30%, we applied k-nearest neighbors (KNN) imputation\u003csup\u003e18\u003c/sup\u003e. Outliers were handled using winsorization (capping): values exceeding the 99th percentile were replaced with the 99th percentile value, and values below the 1st percentile were replaced with the 1st percentile value. Additionally, zero-variance and near-zero-variance variables were excluded. Multicollinearity was assessed to ensure no high correlations existed among the remaining variables. For the external validation phase using the MIMIC-IV and MIMIC-III databases, we adopted a complete-case analysis approach, excluding samples with any missing values to ensure robustness and comparability across datasets.\u003c/p\u003e\n\u003cp\u003e2.6 Association Analysis Methods\u003c/p\u003e\n\u003cp\u003eThis study employed multiple analytical approaches to examine the relationship between malignancies and patient outcomes. Initially, propensity score matching (PSM) \u003csup\u003e19\u003c/sup\u003e was performed to adjust for confounding factors using covariate balancing propensity score (CBPS) estimation \u003csup\u003e20\u003c/sup\u003e. Propensity scores were calculated via CBPS, followed by 1:2 nearest-neighbor matching with a caliper width of 0.1. The effectiveness of PSM was evaluated by calculating standardized mean differences (SMDs), with successful balance defined as all post-matching SMDs \u0026lt;0.1. Successful balance was confirmed when all post-matching standardized mean differences (SMDs) of covariates were \u0026lt;0.1. Weighted logistic regression models were then constructed using the matched sample to estimate effect sizes. To verify result robustness, supplementary analyses were conducted using multivariable logistic regression and IPTW. The multivariable model incorporated all baseline covariates to examine the significance of malignancy-outcome associations. IPTW analysis employed stabilized weights derived from propensity scores, with satisfactory covariate balance achieved after weighting. Consistent findings across PSM, multivariable regression, and IPTW analyses (when regression results from weighted data aligned with both PSM and conventional multivariable results) further reinforced the reliability of our conclusions. This methodological triangulation strategy enhanced the reliability of our observational analyses while maintaining appropriate interpretation boundaries.\u003c/p\u003e\n\u003cp\u003e2.7 Model development and evaluation\u003c/p\u003e\n\u003cp\u003eThis study employed ML approaches to develop predictive models, initially partitioning the eICU database into training and internal validation sets at a 7:3 ratio while addressing class imbalance in the training set using the ROSE algorithm. To prevent overfitting, key variables were selected through the Boruta feature selection method. Subsequently, five ML algorithms—including eXtreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost), k-Nearest Neighbors (KNN), Naïve Bayes (NB), and Logistic Regression (LR)—were implemented, with hyperparameter optimization achieved via 10-fold cross-validation.\u003c/p\u003e\n\u003cp\u003eModel performance was comprehensively evaluated using multiple metrics, including receiver operating characteristic (ROC) curves, Brier score, accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1-score, across both internal validation and two external validation datasets to ensure predictive accuracy and generalizability. Upon completion of model development, we will employ the SHAP method for dual-perspective interpretability analysis: First, global interpretation will be achieved by calculating mean SHAP values across the test cohort to identify the most influential clinical features affecting prediction outcomes; second, individualized SHAP force plots will be generated for both typical cases and special scenarios to visually demonstrate how each feature contributes to prediction results for specific cases. All visualization outputs will be optimized according to clinicians' cognitive characteristics to ensure clinical interpretability and practical utility of the model predictions. This dual-analysis strategy not only validates the biological plausibility of the model at the population level but also provides transparent decision-making references at the individual level\u003csup\u003e21\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFurthermore, a web-based application was developed to facilitate clinical implementation of the final model. All statistical analyses were conducted using R software (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria), with two-sided *p*-values \u0026lt; 0.05 considered statistically significant.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cp\u003eFigure 1 illustrates the technical flowchart of this study, with the left panel depicting the association analysis component. Initially, we identified 28,449 HF patients from the eICU database using ICD codes. After applying stringent inclusion and exclusion criteria, 21,636 patients were enrolled in the study cohort, including 18,239 without a history of malignant tumors and 3,397 with comorbid malignancies. Following association analysis, the 3,397 malignancy-complicated HF patients were randomly split into a training set (2,368 patients for ML model development) and an internal validation set at a 7:3 ratio. To further evaluate model performance, external validation cohorts comprising 1,817 and 539 patients were additionally recruited from the MIMIC-IV and MIMIC-III databases, respectively.\u003c/p\u003e\n\u003cp\u003e3.1 Baseline characteristics\u003c/p\u003e\n\u003cp\u003eThe baseline characteristics before PSM are presented in Table 1. Patients with malignancy exhibited significantly higher disease severity on ICU admission, as reflected by SAPS-II scores (median 35 (IQR: 28–44) vs. 32 (24–41)). They were also older (median age 75 (67–83) vs. 71 (61–81)) and had a higher prevalence of atrial fibrillation (45% vs. 39%) and chronic obstructive pulmonary disease (38% vs. 34%). Laboratory tests differences included lower hemoglobin levels (10.2 g/dL (8.9–11.5) vs. 10.7 (9.3–12.1)) and elevated RDW (15.9 (14.7–17.4) vs. 15.5 (14.5–17.0)). No significant differences were observed in mechanical ventilation (46% vs. 50%) or vasopressor use (20% vs. 20%) within the first 24 hours of ICU admission.\u003c/p\u003e\n\u003cp\u003e3.2 Association Analysis Results\u003c/p\u003e\n\u003cp\u003eHere are the results of our association analysis using PSM and weighted logistic regression. We performed 1:2 PSM with a caliper set to ensure SMD \u0026lt;0.1 for all covariates, achieving good balance between the malignancy and non-malignancy groups in the matched cohort (Table 1, Figure2-A). The weighted logistic regression model, adjusting for all baseline characteristics including demographics, severity scores, comorbidities, vital signs, and laboratory values, demonstrated that malignancy was significantly associated with worse hospital outcomes (OR 1.14, 95% CI 1.02-1.26, p=0.0177). This analysis accounted for the matched nature of the data through robust variance estimation using subclass clustering, and the quasibinomial family was used to address potential overdispersion. The results suggest that malignancy independently predicts poorer outcomes in this critically ill population even after rigorous adjustment for potential confounders through both matching and regression techniques.\u003c/p\u003e\n\u003cp\u003eTo further validate our findings, we conducted sensitivity analyses using IPTW and multivariate logistic regression. The IPTW approach achieved excellent covariate balance (all SMD \u0026lt;0.1) as demonstrated in the Love plot (Figure2-B), and the weighted analysis similarly showed that malignancy was associated with increased mortality risk (OR 1.16, 95% CI 1.03-1.30, p=0.0132). The traditional multivariate logistic regression model, adjusting for all baseline covariates, yielded consistent results (OR 1.20, 95% CI 1.07-1.35, p=0.00224). Figure 3 presents the forest plot comparing these three association analysis methods (PSM, IPTW, and multivariate logistic regression), demonstrating robust and consistent effect estimates across different analytical approaches. The concordance of results from these complementary methods strengthens the evidence for an independent association between malignancy and poorer outcomes in critically ill patients.\u003c/p\u003e\n\u003cp\u003e3.3 Machine Learning Modeling – Feature Selection\u003c/p\u003e\n\u003cp\u003eUsing the Boruta algorithm for all-cause mortality prediction in HF patients with malignancy, we identified 19 key features from comprehensive clinical variables（Figure 4）. The confirmed significant features included illness severity scores (SAPS-II), critical interventions (mechanical ventilation, sedatives, vasopressors), comorbidities (atrial fibrillation), vital signs (mean arterial pressure, heart rate, temperature, respiratory rate, peripheral oxygen saturation), and laboratory tests (bicarbonate, creatinine, blood urea nitrogen, white blood cell count, hemoglobin, platelet count, and electrolytes including chloride, sodium, and potassium). \u003c/p\u003e\n\u003cp\u003e3.4 Model Construction and Comparative Performance Evaluation\u003c/p\u003e\n\u003cp\u003eDuring the model construction phase, we employed five ML algorithmsto develop predictive models. The eICU dataset was partitioned into training and internal validation sets at a 7:3 ratio. To address class imbalance in the training set, we applied the Random Over-Sampling Examples (ROSE) algorithm for data resampling. Model training was optimized through a 10-fold cross-validation approach combined with grid search to systematically tune hyperparameters, ensuring robust performance and generalizability.\u003c/p\u003e\n\u003cp\u003eWe comprehensively evaluated model performance using multiple metrics, conducting internal validation with the partitioned eICU dataset and external validation with both MIMIC-IV and MIMIC-III databases. As illustrated in Figure 5, the ROC curves demonstrate distinct performance patterns across the training set (A), internal validation set (B), MIMIC-IV external validation set (C), and MIMIC-III external validation set (D). While XGBoost achieved superior AUC in the training set, its performance significantly declined across all validation sets, indicating evident overfitting. In contrast, AdaBoost exhibited the most robust and consistent performance across all datasets, particularly maintaining superior predictive accuracy during external validation, suggesting better generalizability for clinical application.\u003c/p\u003e\n\u003cp\u003eAs demonstrated in Table 2, the AdaBoost model demonstrated superior performance in predicting mortality among HF patients across multiple datasets. In external validation (MIMIC-III/IV), AdaBoost achieved the best balance between calibration (Brier Score: 0.187–0.214) and discrimination (Accuracy: 0.672–0.720), outperforming LR and KNN. It exhibited high sensitivity (0.745–0.814) to minimize false negatives and the highest PPV (0.829–0.902) for reliable positive predictions. While XGBoost showed overfitting (F1 drop from 0.983 to 0.853 in validation), AdaBoost maintained robust generalization (F1: 0.810–0.856). Its stability across datasets supports clinical deployment for mortality risk stratification.\u003c/p\u003e\n\u003cp\u003e3.5 SHAP -Based Interpretability Assessment\u003c/p\u003e\n\u003cp\u003eTo enhance the model's interpretability, we performed SHAP analysis on the AdaBoost model. Figure 6 presents the global interpretation plots from SHAP analysis, where Figure 6-A displays the beeswarm plot with model features on the y-axis and their corresponding SHAP values on the x-axis. In the beeswarm plot, yellow dots indicate strong positive contributions to predictions while purple dots represent negative or minimal impacts, with wider horizontal distributions reflecting greater variability in feature effects. Figure 6-B shows the feature importance plot, which ranks variables by their mean absolute SHAP values, providing a comprehensive visualization of each feature's global impact on model outputs. As illustrated in Figure 6, the top five most influential predictive variables were identified as the SAPS-II score (Simplified Acute Physiology Score II), requirement for mechanical ventilation, heart rate, body temperature, and respiratory rate, which collectively demonstrated the strongest associations with clinical outcomes in our predictive model.\u003c/p\u003e\n\u003cp\u003eFigure 7 presents the local interpretability analysis using SHAP force plots, illustrating the model's prediction process for an individual patient. The final prediction output (f(x)=1) demonstrates that negative cumulative effects lowered the predicted value below the baseline, classifying this patient as a survivor. SHAP value analysis identified SAPS-II score, ventilation status, and SpO2 levels as the primary determinants driving this prediction, with each factor's relative contribution quantitatively visualized in the force plot representation.\u003c/p\u003e\n\u003cp\u003e3.6 Interactive Model Visualization for Clinical Decision Support\u003c/p\u003e\n\u003cp\u003eTo facilitate clinical application, we developed a user-friendly web-based calculator (available at: https://nanzihan1998.shinyapps.io/Mortality/) that enables real-time mortality risk assessment for HF patients with comorbid malignancy.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis study comprises two integrated components: association analysis and ML.\u003c/p\u003e\u003cp\u003eThe first component systematically employs three distinct association analysis methodologies to conclusively demonstrate that HF patients with comorbid malignancy experience significantly worse clinical outcomes. Three complementary association analysis approaches consistently demonstrated that malignancy significantly increased in-hospital mortality among ICU-admitted HF patients: PSM (OR 1.14, 95% CI [1.02\u0026ndash;1.26]), IPTW (OR 1.16, 95% CI [1.03\u0026ndash;1.30]), and multivariate logistic regression (OR 1.20, 95% CI [1.07\u0026ndash;1.35]). The concordant results across all methods robustly confirm the detrimental impact of malignancy on clinical outcomes in this vulnerable population.\u003c/p\u003e\u003cp\u003eThe poorer prognosis observed in patients with comorbid malignancies stems from multiple factors, with treatment-related cardiotoxicity representing one of the significant contributors. Anthracyclines, cornerstone agents for various solid and hematologic malignancies\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, induce dose-dependent cardiomyocyte damage through topoisomerase 2β inhibition, activating cell death pathways and impairing mitochondrial biogenesis \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Other chemotherapeutic agents including 5-fluorouracil\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, cyclophosphamide\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, and paclitaxel\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e similarly demonstrate cardiotoxic effects. Beyond chemotherapy, immune checkpoint inhibitors (ICIs) represent a breakthrough in cancer therapy by blocking PD-1/PD-L1 or CTLA-4 to restore antitumor immunity\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. However, ICIs frequently induce inflammatory toxicities termed immune-related adverse events (irAEs) \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. Unlike chemotherapy-induced organ damage, irAEs exhibit unpredictable onset and severity \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. Cardiac involvement occurs in 1.14% of ICI recipients, increasing to 2.4% with combination therapy\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. ICI-associated cardiotoxicity correlates with elevated mortality and often necessitates treatment interruption, requiring high-dose corticosteroids for management\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. In addition to treatment-related cardiotoxicity, cardiac metastasis - particularly to the myocardium or pericardium - can directly compromise cardiac structure and function, exacerbating HF severity. Furthermore, the metabolic demands of rapidly proliferating tumor cells induce a catabolic state characterized by hypoalbuminemia and anemia. This tumor-associated malnutrition exacerbates myocardial energy deficiency in HF patients, whose compromised cardiac output already limits nutrient delivery. The resultant impairment of cardiac repair mechanisms creates a vicious cycle that accelerates disease progression.\u003c/p\u003e\u003cp\u003eThe second component focuses specifically on HF patients with comorbid malignancy, establishing a ML-based predictive model for in-hospital mortality among this high-risk population admitted to ICU. The predictive model was developed using five ML algorithms incorporating 19 carefully selected clinical variables from the first 24 hours of ICU admission. Comparative evaluation demonstrated AdaBoost's superior performance, with Youden's index identifying patients having a predicted probability\u0026thinsp;\u0026gt;\u0026thinsp;0.483 as high-risk for in-hospital mortality. To enhance model credibility, SHAP analysis was employed for comprehensive interpretation - the summary plot provided global visualization of all 19 risk factors' contributions, while force plots enabled individualized prediction explanations for clinical implementation.\u003c/p\u003e\u003cp\u003eThe SAPS-II score emerged as the most significant predictive variable in our model\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. As a well-validated scoring system specifically designed to assess mortality risk in critically ill patients, SAPS-II provide a comprehensive evaluation of illness severity and prognostic outcomes. Its widespread adoption in ICUs and relative ease of acquisition further enhance its clinical utility for risk stratification. Mechanical ventilation requirement ranked as the second most important predictor, with the beeswarm plot demonstrating significantly worse outcomes in ventilated patients. In HF, elevated left ventricular filling pressures frequently lead to alveolar pulmonary edema, impairing oxygenation and ventilation - a pathophysiological cascade culminating in respiratory failure\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Substantial evidence indicates that HF patients complicated by respiratory failure exhibit poorer prognoses, warranting prompt positive-pressure mechanical ventilation to improve clinical outcomes\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. Our study further identified elevated heart rate as an independent predictor of poorer outcomes, a finding consistent with prior evidence. For patients with tachycardia, pharmacological interventions including β-blockers and ivabradine may be considered to attenuate sympathetic overactivation, thereby potentially improving clinical prognosis\u003csup\u003e\u003cspan additionalcitationids=\"CR36\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Notably, observations in heart transplant recipients have also demonstrated a significant correlation between elevated heart rate and deterioration of cardiac function, reinforcing the pathophysiological relevance of heart rate modulation in cardiac patients\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Hypothermia has been consistently associated with adverse clinical outcomes in HF patients. This association may stem from severely compromised cardiac output in advanced disease stages, leading to systemic hypoperfusion, impaired tissue oxygenation, and consequent thermoregulatory failure - a maladaptive cascade that exacerbates end-organ dysfunction. Our model additionally incorporated clinically relevant variables from the following categories: comorbidities (atrial fibrillation); critical interventions within the first 24 hours of ICU admission (vasoactive drugs and sedatives); vital signs (respiratory rate, peripheral oxygen saturation, mean arterial pressure); and laboratory tests (white blood cell count, blood urea nitrogen, creatinine, hemoglobin, bicarbonate, chloride, sodium, potassium, and platelet count).\u003c/p\u003e\u003cp\u003eThis study possesses several notable merits. First, it represents the first investigation to systematically demonstrate worse prognosis in HF patients with comorbid malignancy using three distinct association analysis methodologies. Second, we developed a ML-based predictive model for in-hospital mortality among ICU-admitted HF patients with malignancy, incorporating SHAP analysis to enhance model interpretability and facilitate clinical translation of artificial intelligence technology. Third, the visualization of our model enables user-friendly clinical application, supporting early mortality risk assessment to inform clinical decision-making.\u003c/p\u003e\u003cp\u003eSeveral limitations should be acknowledged. As a multicenter retrospective study, inherent issues of missing data and outliers exist, though we employed rigorous statistical methods to maximize data validity. Our model currently utilizes only cross-sectional data from the first ICU day; future incorporation of longitudinal data may improve predictive accuracy. Additionally, the exclusion of novel biomarkers such as Growth Differentiation Factor-15, mid-regional proatrial natriuretic peptide, and ceruloplasmin may limit predictive performance, suggesting an avenue for model refinement\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eThrough robust association analysis, we demonstrated that malignancy significantly worsens outcomes in HF patients. We developed and validated a ML model to predict in-hospital mortality risk in this high-risk population. SHAP analysis enhanced model interpretability, enabling clinicians to identify key mortality predictors and facilitate timely interventions to improve patient prognosis.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAdaBoost: Adaptive Boosting; CBPS: covariate balancing propensity score; CRRT: continuous renal replacement therapy; eICU: eICU Collaborative Research Database; HF: heart failure; ICD: International Classification of Diseases; ICI: immune checkpoint inhibitor; ICU: Intensive Care Unit; IPTW: inverse probability treatment weighting; KNN: k-nearest neighbors; LR: Logistic Regression; MIMIC: Medical Information Mart for Intensive Care; ML: machine learning; NB: Naïve Bayes; PSM: propensity score matching; RDW: red cell distribution width; ROSE: Random Over-Sampling Examples; SAPS-II: Simplified Acute Physiology Score II; SHAP: SHapley Additive exPlanations; SMD: standardized mean difference; SpO2: peripheral oxygen saturation; STROBE: STrengthening the Reporting of OBservational studies in Epidemiology; TRIPOD: Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis; XGBoost: eXtreme Gradient Boosting.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to express our sincere gratitude to the official operators of the MIMIC and eICU databases for providing access to these valuable clinical datasets. This study was supported by the Health Commission of Hebei Province and the Department of Science and Technology of Hebei Province. Their support was instrumental in facilitating this research endeavor.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTao Zhang and Zihan Nan created the study protocol and wrote the first manuscript draft. Zihan Nan performed all data collection and statistical analyses. Shaohan Guo conceived the study and critically revised the manuscript. Shiju Yang assisted with the study design. Leying Li assisted with manuscript editing. Yifei Zhang assisted with the interpretation of statistical methods. Yan Xin assisted with manuscript revision and data validation. Zhenjie Hu and Congcong Zhao supervised the study, contributed to data interpretation, and finalized the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Medical Science Research Project of Hebei (Grant No. 20250673) and the S\u0026amp;T Program of Hebei (Grant No. 223777104D).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data extracted in this study are available from the corresponding authors upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe eICU database and MIMIC-III/IV databases were also approved by the Massachusetts Institute of Technology (Cambridge, MA) and Beth Israel Deaconess Medical Center (Boston, MA), and consent was obtained for the original data collection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflicts of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eKoene, R. J., Prizment, A. E., Blaes, A. \u0026amp; Konety, S. H. Shared Risk Factors in Cardiovascular Disease and Cancer. \u003cem\u003eCirculation\u003c/em\u003e \u003cstrong\u003e133\u003c/strong\u003e, 1104-1114, doi:10.1161/CIRCULATIONAHA.115.020406 (2016).\u003c/li\u003e\n\u003cli\u003eKhan, M. S.\u003cem\u003e et al.\u003c/em\u003e Global epidemiology of heart failure. \u003cem\u003eNat Rev Cardiol\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 717-734, doi:10.1038/s41569-024-01046-6 (2024).\u003c/li\u003e\n\u003cli\u003eWang, H.\u003cem\u003e et al.\u003c/em\u003e Mortality in patients admitted to hospital with heart failure in China: a nationwide Cardiovascular Association Database-Heart Failure Centre Registry cohort study. \u003cem\u003eLancet Glob Health\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e611-e622, doi:10.1016/S2214-109X(23)00605-8 (2024).\u003c/li\u003e\n\u003cli\u003eWu, Z., Xia, F. \u0026amp; Lin, R. Global burden of cancer and associated risk factors in 204 countries and territories, 1980-2021: a systematic analysis for the GBD 2021. \u003cem\u003eJ Hematol Oncol\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 119, doi:10.1186/s13045-024-01640-8 (2024).\u003c/li\u003e\n\u003cli\u003eNohria, A. Double Jeopardy: Cancer and Heart Failure. \u003cem\u003eJ Card Fail\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 522-523, doi:10.1016/j.cardfail.2019.05.002 (2019).\u003c/li\u003e\n\u003cli\u003eTuzovic, M.\u003cem\u003e et al.\u003c/em\u003e National Outcomes in Hospitalized Patients With Cancer and Comorbid Heart Failure. \u003cem\u003eJ Card Fail\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 516-521, doi:10.1016/j.cardfail.2019.02.007 (2019).\u003c/li\u003e\n\u003cli\u003eLi, J.\u003cem\u003e et al.\u003c/em\u003e Predicting Mortality in Intensive Care Unit Patients With Heart Failure Using an Interpretable Machine Learning Model: Retrospective Cohort Study. \u003cem\u003eJ Med Internet Res\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, e38082, doi:10.2196/38082 (2022).\u003c/li\u003e\n\u003cli\u003ePeng, S.\u003cem\u003e et al.\u003c/em\u003e Interpretable machine learning for 28-day all-cause in-hospital mortality prediction in critically ill patients with heart failure combined with hypertension: A retrospective cohort study based on medical information mart for intensive care database-IV and eICU databases. \u003cem\u003eFront Cardiovasc Med\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 994359, doi:10.3389/fcvm.2022.994359 (2022).\u003c/li\u003e\n\u003cli\u003eVinter, N.\u003cem\u003e et al.\u003c/em\u003e Development and validation of prediction models for incident atrial fibrillation in heart failure. \u003cem\u003eOpen Heart\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, doi:10.1136/openhrt-2022-002169 (2023).\u003c/li\u003e\n\u003cli\u003eHounye, A. H., Xiong, L. \u0026amp; Hou, M. Integrated explainable machine learning and multi-omics analysis for survival prediction in cancer with immunotherapy response. \u003cem\u003eApoptosis\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 364-388, doi:10.1007/s10495-024-02050-4 (2024).\u003c/li\u003e\n\u003cli\u003eMontazeri, M., Montazeri, M., Montazeri, M. \u0026amp; Beigzadeh, A. Machine learning models in breast cancer survival prediction. \u003cem\u003eTechnology and Health Care\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 31-42, doi:10.3233/thc-151071 (2016).\u003c/li\u003e\n\u003cli\u003eLundberg, S. M. \u0026amp; Lee, S.-I. A unified approach to interpreting model predictions. \u003cem\u003eAdvances in neural information processing systems\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e (2017).\u003c/li\u003e\n\u003cli\u003eVandenbroucke, J. P.\u003cem\u003e et al.\u003c/em\u003e Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. \u003cem\u003ePLoS Med\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, e297, doi:10.1371/journal.pmed.0040297 (2007).\u003c/li\u003e\n\u003cli\u003eCollins, G. S., Reitsma, J. B., Altman, D. G. \u0026amp; Moons, K. G. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. \u003cem\u003eBMJ\u003c/em\u003e \u003cstrong\u003e350\u003c/strong\u003e, g7594, doi:10.1136/bmj.g7594 (2015).\u003c/li\u003e\n\u003cli\u003ePollard, T. J.\u003cem\u003e et al.\u003c/em\u003e The eICU Collaborative Research Database, a freely available multi-center database for critical care research. \u003cem\u003eSci Data\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 180178, doi:10.1038/sdata.2018.178 (2018).\u003c/li\u003e\n\u003cli\u003eJohnson, A. E. W.\u003cem\u003e et al.\u003c/em\u003e MIMIC-IV, a freely accessible electronic health record dataset. \u003cem\u003eSci Data\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 1, doi:10.1038/s41597-022-01899-x (2023).\u003c/li\u003e\n\u003cli\u003eJohnson, A. E.\u003cem\u003e et al.\u003c/em\u003e MIMIC-III, a freely accessible critical care database. \u003cem\u003eSci Data\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 160035, doi:10.1038/sdata.2016.35 (2016).\u003c/li\u003e\n\u003cli\u003ePujianto, U., Wibawa, A. P. \u0026amp; Akbar, M. I. in \u003cem\u003e2019 5th International Conference on Science in Information Technology (ICSITech).\u003c/em\u003e 83-88 (IEEE).\u003c/li\u003e\n\u003cli\u003eHaukoos, J. S. \u0026amp; Lewis, R. J. The propensity score. \u003cem\u003eJama\u003c/em\u003e \u003cstrong\u003e314\u003c/strong\u003e, 1637-1638 (2015).\u003c/li\u003e\n\u003cli\u003eImai, K. \u0026amp; Ratkovic, M. Covariate balancing propensity score. \u003cem\u003eJournal of the Royal Statistical Society Series B: Statistical Methodology\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, 243-263 (2014).\u003c/li\u003e\n\u003cli\u003eLundberg, S. M.\u003cem\u003e et al.\u003c/em\u003e From local explanations to global understanding with explainable AI for trees. \u003cem\u003eNature Machine Intelligence\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 56-67, doi:10.1038/s42256-019-0138-9 (2020).\u003c/li\u003e\n\u003cli\u003eBloom, M. W.\u003cem\u003e et al.\u003c/em\u003e Cardio-Oncology and Heart Failure: a Scientific Statement From the Heart Failure Society of America. \u003cem\u003eJ Card Fail\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 415-455, doi:10.1016/j.cardfail.2024.08.045 (2025).\u003c/li\u003e\n\u003cli\u003eHenriksen, P. A. Anthracycline cardiotoxicity: an update on mechanisms, monitoring and prevention. \u003cem\u003eHeart\u003c/em\u003e \u003cstrong\u003e104\u003c/strong\u003e, 971-977, doi:10.1136/heartjnl-2017-312103 (2018).\u003c/li\u003e\n\u003cli\u003eAgunbiade, T. A., Zaghlol, R. Y. \u0026amp; Barac, A. Heart Failure in Relation to Anthracyclines and Other Chemotherapies. \u003cem\u003eMethodist Debakey Cardiovasc J\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 243-249, doi:10.14797/mdcj-15-4-243 (2019).\u003c/li\u003e\n\u003cli\u003eGradishar, W. J. \u0026amp; Vokes, E. E. 5-Fluorouracil cardiotoxicity: a critical review. \u003cem\u003eAnn Oncol\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 409-414, doi:10.1093/oxfordjournals.annonc.a057793 (1990).\u003c/li\u003e\n\u003cli\u003eIqubal, A.\u003cem\u003e et al.\u003c/em\u003e Molecular mechanism involved in cyclophosphamide-induced cardiotoxicity: Old drug with a new vision. \u003cem\u003eLife Sci\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, 112-131, doi:10.1016/j.lfs.2018.12.018 (2019).\u003c/li\u003e\n\u003cli\u003ePerez, E. A. Paclitaxel and cardiotoxicity. \u003cem\u003eJ Clin Oncol\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 3481-3482, doi:10.1200/JCO.1998.16.11.3481 (1998).\u003c/li\u003e\n\u003cli\u003eArafat Hossain, M. A comprehensive review of immune checkpoint inhibitors for cancer treatment. \u003cem\u003eInt Immunopharmacol\u003c/em\u003e \u003cstrong\u003e143\u003c/strong\u003e, 113365, doi:10.1016/j.intimp.2024.113365 (2024).\u003c/li\u003e\n\u003cli\u003eWang, S. J., Dougan, S. K. \u0026amp; Dougan, M. Immune mechanisms of toxicity from checkpoint inhibitors. \u003cem\u003eTrends Cancer\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 543-553, doi:10.1016/j.trecan.2023.04.002 (2023).\u003c/li\u003e\n\u003cli\u003eMahmood, S. S.\u003cem\u003e et al.\u003c/em\u003e Myocarditis in Patients Treated With Immune Checkpoint Inhibitors. \u003cem\u003eJ Am Coll Cardiol\u003c/em\u003e \u003cstrong\u003e71\u003c/strong\u003e, 1755-1764, doi:10.1016/j.jacc.2018.02.037 (2018).\u003c/li\u003e\n\u003cli\u003ePatel, R. P.\u003cem\u003e et al.\u003c/em\u003e Cardiotoxicity of Immune Checkpoint Inhibitors. \u003cem\u003eCurr Oncol Rep\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 79, doi:10.1007/s11912-021-01070-6 (2021).\u003c/li\u003e\n\u003cli\u003eKeegan, M. T., Gajic, O. \u0026amp; Afessa, B. Severity of illness scoring systems in the intensive care unit. \u003cem\u003eCrit Care Med\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 163-169, doi:10.1097/CCM.0b013e3181f96f81 (2011).\u003c/li\u003e\n\u003cli\u003eAlviar, C. L.\u003cem\u003e et al.\u003c/em\u003e Positive Pressure Ventilation in Cardiogenic Shock: Review of the Evidence and Practical Advice for Patients With Mechanical Circulatory Support. \u003cem\u003eCan J Cardiol\u003c/em\u003e \u003cstrong\u003e36\u003c/strong\u003e, 300-312, doi:10.1016/j.cjca.2019.11.038 (2020).\u003c/li\u003e\n\u003cli\u003eTavazzi, G. Mechanical ventilation in cardiogenic shock. \u003cem\u003eCurr Opin Crit Care\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 447-453, doi:10.1097/MCC.0000000000000836 (2021).\u003c/li\u003e\n\u003cli\u003eKapoor, J. R. \u0026amp; Heidenreich, P. A. Role of heart rate as a marker and mediator of poor outcome for patients with heart failure. \u003cem\u003eCurr Heart Fail Rep\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 133-138, doi:10.1007/s11897-012-0086-8 (2012).\u003c/li\u003e\n\u003cli\u003eSciatti, E., Vizzardi, E., Bonadei, I., Dallapellegrina, L. \u0026amp; Carubelli, V. The role of heart rate and ivabradine in acute heart failure. \u003cem\u003eMonaldi Arch Chest Dis\u003c/em\u003e \u003cstrong\u003e89\u003c/strong\u003e, doi:10.4081/monaldi.2019.1091 (2019).\u003c/li\u003e\n\u003cli\u003eKubrychtova, V.\u003cem\u003e et al.\u003c/em\u003e Heart rate recovery and prognosis in heart failure patients. \u003cem\u003eEur J Appl Physiol\u003c/em\u003e \u003cstrong\u003e105\u003c/strong\u003e, 37-45, doi:10.1007/s00421-008-0870-z (2009).\u003c/li\u003e\n\u003cli\u003eLiebo, M.\u003cem\u003e et al.\u003c/em\u003e Elevated Heart Rate Following Heart Transplantation Is Associated With Increased Graft Vasculopathy and Mortality. \u003cem\u003eJ Card Fail\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 249-256, doi:10.1016/j.cardfail.2019.01.009 (2019).\u003c/li\u003e\n\u003cli\u003eSavic-Radojevic, A.\u003cem\u003e et al.\u003c/em\u003e Novel Biomarkers of Heart Failure. \u003cem\u003eAdv Clin Chem\u003c/em\u003e \u003cstrong\u003e79\u003c/strong\u003e, 93-152, doi:10.1016/bs.acc.2016.09.002 (2017).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 and 2 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"heart failure, malignancy, in-hospital mortality, machine learning, interpretable AI, predictive modeling","lastPublishedDoi":"10.21203/rs.3.rs-7179137/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7179137/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eHeart failure (HF) and malignancy represent two major global health burdens that frequently coexist and lead to poor clinical outcomes. However, the specific impact of malignancy on in-hospital mortality in HF patients remains incompletely understood, and reliable predictive tools specifically for HF patients with comorbid malignancy are currently lacking.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eThis multicenter retrospective study analyzed data from the eICU Collaborative Research Database (eICU database), Medical Information Mart for Intensive Care IV (MIMIC-IV) databases, and Medical Information Mart for Intensive Care III (MIMIC-III) databases, including 21,636 HF patients (3,397 with malignancy). We employed three analytical approaches: propensity score matching (PSM), inverse probability treatment weighting (IPTW), and multivariable logistic regression to assess malignancy-associated mortality risk. For predictive modeling, five machine learning algorithms were trained on the eICU database (70% training, 30% internal validation) and externally validated using both MIMIC-IV and MIMIC-III datasets.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eAll three analytical methods (PSM, IPTW, and multivariable regression) yielded highly consistent results, demonstrating that malignancy significantly increased in-hospital mortality risk (PSM: OR 1.14, 95% CI 1.02\u0026ndash;1.26; IPTW: OR 1.16, 95% CI 1.03\u0026ndash;1.30; multivariable regression: OR 1.20, 95% CI 1.07\u0026ndash;1.35). The AdaBoost model, developed using 19 key predictive variables selected by the Boruta algorithm, demonstrated excellent performance with a training set AUC of 0.849 and internal validation AUC of 0.740, while maintaining good discriminative ability in external validation (MIMIC-IV: AUC 0.739; MIMIC-III: AUC 0.699). To enhance model interpretability, SHapley Additive exPlanations analysis revealed the top five predictive variables: Simplified Acute Physiology Score II, mechanical ventilation requirement, heart rate, body temperature, and respiratory rate. For clinical implementation, we developed a web-based calculator (available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://nanzihan1998.shinyapps.io/Mortality/\u003c/span\u003e\u003cspan address=\"https://nanzihan1998.shinyapps.io/Mortality/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to facilitate real-time mortality risk assessment.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eMalignancy independently worsens outcomes in HF patients. Our interpretable machine learning model incorporating multiple clinically relevant predictors provides accurate mortality risk stratification, facilitating personalized clinical decision-making for this high-risk population. Future studies should incorporate longitudinal data and novel biomarkers to further improve predictive performance.\u003c/p\u003e","manuscriptTitle":"Multicenter Machine Learning Model for Assessing the Impact of Malignancy on In-Hospital Mortality in Heart Failure Patients: A Clinical Decision Support System with Interpretable Artificial Intelligence","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-21 10:25:37","doi":"10.21203/rs.3.rs-7179137/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-04T16:53:49+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-24T09:11:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"246377234376188366546263993189164379159","date":"2025-09-10T14:00:23+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"70386932136913223887417914753242408988","date":"2025-09-09T07:37:00+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-31T19:16:17+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"90742084949470884844229017326610710122","date":"2025-08-16T14:37:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"148487497305395365358077433532635770571","date":"2025-08-13T06:42:01+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-13T05:25:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-13T05:15:23+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-07-28T12:11:09+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-24T14:21:30+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-07-24T14:17:47+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2fd8e2f0-79c4-4dee-8097-c6f2674f3a18","owner":[],"postedDate":"August 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":53287466,"name":"Health sciences/Biomarkers"},{"id":53287467,"name":"Biological sciences/Cancer"},{"id":53287468,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":53287469,"name":"Health sciences/Diseases"},{"id":53287470,"name":"Health sciences/Medical research"},{"id":53287471,"name":"Health sciences/Oncology"},{"id":53287472,"name":"Health sciences/Risk factors"}],"tags":[],"updatedAt":"2026-05-06T04:24:43+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-21 10:25:37","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7179137","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7179137","identity":"rs-7179137","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}