Improving risk prediction in heart failure using machine learning

Machine learning-based 28-day mortality prediction model for elderly neurocritically Ill patients

BACKGROUND

The growing population of elderly neurocritically ill patients highlights the need for effective prognosis prediction tools. This study aims to develop and validate machine learning (ML) models for predicting 28-day mortality in intensive care units (ICUs).

METHODS

Data were extracted from the Medical Information Mart for Intensive Care IV(MIMIC-IV) database, focusing on elderly neurocritical ill patients with ICU stays ≥ 24 h. The cohort was split into 70 % for training and 30 % for internal validation. We analyzed 58 variables, including demographics, vital signs, medications, lab results, comorbidities, and medical scores, using Lasso regression to identify predictors of 28-day mortality. Seven ML algorithms were evaluated, and the best model was validated with data from Guizhou Medical University Affiliated Hospital. A log-rank test was used to assess survival differences in Kaplan-Meier curves. Shapley Additive Explanations (SHAP) were used to interpret the best model, while subgroup analysis identified variations in model performance across different populations.

RESULTS

The study included 1,773 elderly neurocritically ill patients, with a 28-day mortality rate of 28.6 %. The Light Gradient Boosting Machine (LightGBM) outperformed other models, achieving an area under the curve (AUC) of 0.896 in internal validation and 0.812 in external validation. Kaplan-Meier analysis showed that higher LightGBM prediction scores correlated with lower survival probabilities. Key predictors identified through SHAP analysis included partial pressure of arterial carbon dioxide (PaCO2), Acute physiology and chronic health evaluation II (APACHE II), white blood cell count, age, and lactate. The LightGBM model demonstrated consistent performance across various subgroups.

CONCLUSIONS

The LightGBM model effectively predicts 28-day mortality risk in elderly neurocritically ill patients, aiding clinicians in management and resource allocation. Its reliable performance across diverse subgroups underscores its clinical utility.

The Heart of Transformation: Exploring Artificial Intelligence in Cardiovascular Disease

The application of artificial intelligence (AI) and machine learning (ML) in medicine and healthcare has been extensively explored across various areas. AI and ML can revolutionize cardiovascular disease management by significantly enhancing diagnostic accuracy, disease prediction, workflow optimization, and resource utilization. This review summarizes current advancements in AI and ML concerning cardiovascular disease, including their clinical investigation and use in primary cardiac imaging techniques, common cardiovascular disease categories, clinical research, patient care, and outcome prediction. We analyze and discuss commonly used AI and ML models, algorithms, and methodologies, highlighting their roles in improving clinical outcomes while addressing current limitations and future clinical applications. Furthermore, this review emphasizes the transformative potential of AI and ML in cardiovascular practice by improving clinical decision making, reducing human error, enhancing patient monitoring and support, and creating more efficient healthcare workflows for complex cardiovascular conditions.

Prediction of mortality in heart failure by machine learning. Comparison with statistical modeling

BACKGROUND

Assessing the relative performance of machine learning (ML) methods and conventional statistical methods in predicting prognosis in heart failure (HF) still remains a challenging research field.

METHODS

The primary outcome was three-year mortality. The following 5 machine learning approaches were used for modeling: Random Forest (RF), Gradient Boosting, Extreme Gradient Boosting (XGBoost), Support Vector Machine, and Multilayer perceptron. We compared the performance of the best performing ML models to the MAGGIC (Meta-analysis Global Group in Chronic Heart Failure) score and a novel logistic regression model (LRM) developed using the same set of variables used to develop the machine learning models. The performance was determined based on discrimination, calibration, and net benefit.

RESULTS

The XGBoost and the RF were the best performing ML models. The XGBoost provided the highest discrimination (C-statistic: 0.793) and the lowest Brier score (0.178); the RF model had a C-statistic of 0.779 and provided the highest area under the precision-recall curve (0.636). Both models were well calibrated. Both the XGboost and RF models outperformed MAGGIC score. The LRM had a C-statistic of 0.811 and a Brier score of 0.160 and was well calibrated. The XGBoost, RF, and LRM gave a higher net benefit than MAGGIC score; the XGBoost, RF, and logistic regression model gave similar net benefit.

CONCLUSIONS

RF and XGBoost models outperformed MAGGIC in predicting mortality. However, they did not offer any improvement over a logistic regression model built using the same set of covariates considered in the ML modeling.

Revolutionizing Cardiac Care: Artificial Intelligence Applications in Heart Failure Management

Recent advancements in artificial intelligence (AI) have revolutionized the diagnosis, risk assessment, and treatment of heart failure (HF). AI models have demonstrated superior performance in distinguishing healthy individuals from those at risk of congestive HF by analyzing heart rate variability data. In addition, AI clinical decision support systems exhibit high concordance rates with HF experts, enhancing diagnostic precision. For HF with reduced as well as preserved ejection fraction, AI-powered algorithms help detect subtle irregularities in electrocardiograms and other related predictors. AI also aids in predicting HF risk in diabetic patients, using complex data patterns to enhance understanding and management. Moreover, AI technologies help forecast HF-related hospital admissions, enabling timely interventions to reduce readmission rates and improve patient outcomes. Continued innovation and research are crucial to address challenges related to data privacy and ethical considerations and ensure responsible implementation in healthcare.

Enhancing Clinical Decision Making by Predicting Readmission Risk in Patients With Heart Failure Using Machine Learning: Predictive Model Development Study

Background

Patients with heart failure frequently face the possibility of rehospitalization following an initial hospital stay, placing a significant burden on both patients and health care systems. Accurate predictive tools are crucial for guiding clinical decision-making and optimizing patient care. However, the effectiveness of existing models tailored specifically to the Chinese population is still limited.

Objective

This study aimed to formulate a predictive model for assessing the likelihood of readmission among patients diagnosed with heart failure.

Methods

In this study, we analyzed data from 1948 patients with heart failure in a hospital in Sichuan Province between 2016 and 2019. By applying 3 variable selection strategies, 29 relevant variables were identified. Subsequently, we constructed 6 predictive models using different algorithms: logistic regression, support vector machine, gradient boosting machine, Extreme Gradient Boosting, multilayer perception, and graph convolutional networks.

Results

The graph convolutional network model showed the highest prediction accuracy with an area under the receiver operating characteristic curve of 0.831, accuracy of 75%, sensitivity of 52.12%, and specificity of 90.25%.

Conclusions

The model crafted in this study proves its effectiveness in forecasting the likelihood of readmission among patients with heart failure, thus serving as a crucial reference for clinical decision-making.

Construction of Clinical Predictive Models for Heart Failure Detection Using Six Different Machine Learning Algorithms: Identification of Key Clinical Prognostic Features

Purpose

Heart failure (HF) is a clinical syndrome in which structural or functional abnormalities of the heart result in impaired ventricular filling or ejection capacity. In order to improve the adaptability of models to different patient populations and data situations. This study aims to develop predictive models for HF risk using six machine learning algorithms, providing valuable insights into the early assessment and recognition of HF by clinical features.

Patients and Methods

The present study focused on clinical characteristics that significantly differed between groups with left ventricular ejection fractions (LVEF) [≤40% and >40%]. Following the elimination of features with significant missing values, the remaining features were utilized to construct predictive models employing six machine learning algorithms. The optimal model was selected based on various performance metrics, including the area under the curve (AUC), accuracy, precision, recall, and F1 score. Utilizing the optimal model, the significance of clinical features was assessed, and those with importance values exceeding 0.8 were identified as crucial to the study. Finally, a correlation analysis was conducted to examine the relationships between these features and other significant clinical features.

Results

The logistic regression (LR) model was determined to be the optimal machine learning algorithm in this study, achieving an accuracy of 0.64, a precision of 0.45, a recall of 0.72, an F1 score of 0.51, and an AUC of 0.81 in the training set and 0.91 in the testing set. In addition, the analysis of feature importance indicated that blood calcium, angiotensin-converting enzyme inhibitors (ACEI) dosage, mean hemoglobin concentration, and survival duration were critical to the study, each possessing importance values exceeding 0.8. Furthermore, correlation analysis revealed a strong relationship between blood calcium and ionized calcium (|cor|=0.99), as well as a significant association between ACEI dosage (|cor|=0.68) and left ventricular metrics (|cor|=0.58); on the other hand, no correlations were observed between mean hemoglobin levels and other clinical characteristics.

Conclusion

The present study identified LR as the most effective risk prediction model for patients with HF, highlighting blood calcium, ACEI dosage, and mean hemoglobin level as significant predictors. These findings provide significant insights for the clinical prevention and early intervention of HF.

Predicting in-hospital mortality in patients with heart failure combined with atrial fibrillation using stacking ensemble model: an analysis of the medical information mart for intensive care IV (MIMIC-IV)

BACKGROUND

Heart failure (HF) and atrial fibrillation (AF) usually coexist and are associated with a poorer prognosis. This study aimed to develop a model to predict in-hospital mortality in patients with HF combined with AF.

METHODS

Patients with HF and AF were obtained from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database from 2008 to 2019. Feature selection was based on the Mann-Whitney U test and the least absolute shrinkage and selection operator (LASSO) regression model. Random Forest, eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LGBM), K-Nearest Neighbor (KNN) models, and their stacked model (the stacking ensemble model) were established. The area under of the curve (AUC) with 95% confidence interval (CI), sensitivity, specificity, as well as accuracy were applied to assess the performance of the predictive models.

RESULTS

A total of 5,998 patients with HF combined with AF were included, of which 4,198 patients were assigned to the training set and 1,800 to the testing set (7:3). Among these 4,198 patients, 624 (14.86%) died in-hospital and 3,574 (85.14%) survived. Twenty-two features were used to construct the predictive model. Among these four single models, the AUC was 0.747 (95%CI: 0.717-0.777) for the Random Forest model, 0.755 (95%CI: 0.725-0.785) for the XGBoost model, 0.754 (95%CI: 0.724-0.784) for the LGBM model, and 0.746 (95%CI: 0.716-0.776) for the KNN model in the testing set. The stacking ensemble model had the highest AUC compared to the four single models, with AUCs of 0.837 (95%CI: 0.821-0.852) and 0.768 (95%CI: 0.740-0.796) for the training set and testing set, respectively.

CONCLUSION

The stacking ensemble model showed a good predictive effect in predicting in-hospital mortality in patients with HF combined with AF and may provide clinicians with a reference tool for early identification of mortality risk.

A medical disease assisted diagnosis method based on lightweight fuzzy SZGWO-ELM neural network model

The application of neural network model in intelligent diagnosis usually encounters challenges such as continuous adjustment of network parameters and significant cost in training the network facing numerous complex physiological data. To address this challenge, this paper introduces a fuzzy SZGWO-ELM neural network model for medical disease aid diagnosis with fuzzy membership function and ELM network to refine the improved Gray Wolf optimization algorithm. Firstly, the Z-type membership function is introduced as the inertia weight to get a balance for the grey wolf in seeking the optimal solution globally and locally and ensuring fast convergence. Secondly, the S-type membership function is utilized as the adaptive weight to flexibly adjust the grey wolf search step size to facilitate a quick approximation of the optimal solution. Finally, the improved Gray Wolf optimization algorithm is used to optimize the parameters of the ELM neural network model, termed as SZGWO-ELM. This method can eliminate the need for extensive network parameter adjustments and quickly locate the optimal solution to the problem using a lightweight neural network. The performance of the SZGWO is assessed by using metrics like convergence, mean, and standard deviation. Multiple experiments reveal that this method shows superior performance. Furthermore, five publicly accessible medical disease datasets from UCI were conducted to evaluate the performance of SZGWO-ELM network model comparing with different classify model, and the results in terms of precision, sensitivity, specificity and accuracy can achieve 99.52%, 94.14%, 99.26% and 96.08%, respectively, which illustrate that the proposed SZGWO-ELM neural network significantly enhance the model's accuracy, providing better support for doctors in disease diagnosis.

In-hospital bioimpedance-derived total body water predicts short-term cardiovascular mortality and re-hospitalizations in acute decompensated heart failure patients

BACKGROUND

Hospital re-admissions in heart failure (HF) patients are mostly caused by an acute exacerbation of their chronic congestion. Bioimpedance analysis (BIA) has emerged as a promising non-invasive method to assess the volume status in HF. However, its correlation with clinically assessed volume status and its prognostic value in the acute intra-hospital setting remains uncertain.

METHODS AND RESULTS

In this single-center observational study, patients (n = 49) admitted to the cardiology ward for acute decompensated HF (ADHF) underwent a daily BIA-derived volume status assessment. Median hospital stay was 7 (4-10) days. Twenty patients (40%) reached the composite endpoint of cardiovascular mortality or re-hospitalization for HF over 6 months. Patients at discharge displayed improved NYHA class, lower body weight, plasma and blood volume, as well as lower NT-proBNP levels compared to the admission. Compared to patients with total body water (TBW) less than or equal to that predicted by body weight, those with higher relative TBW levels had elevated NT-proBNP and E/e´ (both p < 0.05) at discharge. In the Cox multivariate regression analysis, the BIA-derived delta TBW between admission and discharge showed a 23% risk reduction for each unit increase (HR = 0.776; CI 0.67-0.89; p = 0.0006). In line with this finding, TBW at admission had the highest prediction importance of the combined endpoint for a subgroup of high-risk HF patients (n = 35) in a neural network analysis.

CONCLUSION

In ADHF patients, BIA-derived TBW is associated with the increased risk of HF hospitalization or cardiovascular death over 6 months. The role of BIA for prognostic stratification merits further investigation.

Predictive Analytics in Heart Failure Risk, Readmission, and Mortality Prediction: A Review

Heart failure is a leading cause of death among people worldwide. The cost of treatment can be prohibitive, and early prediction of heart failure would reduce treatment costs to patients and hospitals. Improved readmission prediction would also greatly help hospitals, allowing them to manage their treatment programs and budgets better. This literature review aims to summarize recent studies of predictive analytics models that have been constructed to predict heart failure risk, readmission, and mortality. Random forest, logistic regression, neural networks, and XGBoost were among the most common modeling techniques applied. Most selected studies leveraged structured electronic health record data, including demographics, clinical values, lifestyle, and comorbidities, with some incorporating unstructured clinical notes. Preprocessing through imputation and feature selection were frequently employed in building the predictive analytics models. The reviewed studies exhibit demonstrated promise for predictive analytics in improving early heart failure diagnosis, readmission risk stratification, and mortality prediction. This review study highlights rising research activities and the potential of predictive analytics, especially the implementation of machine learning, in advancing heart failure outcomes. Further rigorous, comprehensive syntheses and head-to-head benchmarking of predictive models are needed to derive robust evidence for clinical adoption.

A Machine Learning-derived Risk Score Improves Prediction of Outcomes After LVAD Implantation: An Analysis of the INTERMACS Database

BACKGROUND

Significant variability in outcomes after left ventricular assist device (LVAD) implantation emphasize the importance of accurately assessing patients' risk before surgery. This study assesses the Machine Learning Assessment of Risk and Early Mortality in Heart Failure (MARKER-HF) mortality risk model, a machine learning-based tool using 8 clinical variables, to predict post-LVAD implantation mortality and its prognostic enhancement over the Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS) profile.

METHODS

Analyzing 25,365 INTERMACS database patients (mean age 56.8 years, 78% male), 5,663 (22.3%) and 19,702 (77.7%) received HeartMate 3 and other types of LVAD, respectively. They were categorized into low, moderate, high, and very high-risk groups based on MARKER-HF score. The outcomes of interest were in-hospital and 1-year postdischarge mortality.

RESULTS

In patients receiving HeartMate 3 devices, 6.2% died during the index hospitalization. In-hospital mortality progressively increased from 4.4% in low-risk to 15.2% in very high-risk groups with MARKER-HF score. MARKER-HF provided additional risk discrimination within each INTERMACS profile. Combining MARKER-HF score and INTERMACS profile identified patients with the lowest (3.5%) and highest in-hospital mortality rates (19.8%). The postdischarge mortality rate at 1 year was 5.8% in this population. In a Cox proportional hazard regression analysis adjusting for both MARKER-HF and INTERMACS profile, only MARKER-HF score (hazard ratio 1.27, 95% confidence interval 1.11-1.46, P < .001) was associated with postdischarge mortality. Similar findings were observed for patients receiving other types of LVADs.

CONCLUSIONS

The MARKER-HF score is a valuable tool for assessing mortality risk in patients with HF undergoing HeartMate 3 and other LVAD implantation. It offers prognostic information beyond that of the INTERMACS profile alone and its use should help in the shared decision-making process for LVAD implantation.

Integrated Approach Using Intuitionistic Fuzzy Multicriteria Decision-Making to Support Classifier Selection for Technology Adoption in Patients with Parkinson Disease: Algorithm Development and Validation

Background

Parkinson disease (PD) is reported to be among the most prevalent neurodegenerative diseases globally, presenting ongoing challenges and increasing burden on health care systems. In an effort to support patients with PD, their carers, and the wider health care sector to manage this incurable condition, the focus has begun to shift away from traditional treatments. One of the most contemporary treatments includes prescribing assistive technologies (ATs), which are viewed as a way to promote independent living and deliver remote care. However, the uptake of these ATs is varied, with some users not ready or willing to accept all forms of AT and others only willing to adopt low-technology solutions. Consequently, to manage both the demands on resources and the efficiency with which ATs are deployed, new approaches are needed to automatically assess or predict a user's likelihood to accept and adopt a particular AT before it is prescribed. Classification algorithms can be used to automatically consider the range of factors impacting AT adoption likelihood, thereby potentially supporting more effective AT allocation. From a computational perspective, different classification algorithms and selection criteria offer various opportunities and challenges to address this need.

Objective

This paper presents a novel hybrid multicriteria decision-making approach to support classifier selection in technology adoption processes involving patients with PD.

Methods

First, the intuitionistic fuzzy analytic hierarchy process (IF-AHP) was implemented to calculate the relative priorities of criteria and subcriteria considering experts' knowledge and uncertainty. Second, the intuitionistic fuzzy decision-making trial and evaluation laboratory (IF-DEMATEL) was applied to evaluate the cause-effect relationships among criteria/subcriteria. Finally, the combined compromise solution (CoCoSo) was used to rank the candidate classifiers based on their capability to model the technology adoption.

Results

We conducted a study involving a mobile smartphone solution to validate the proposed methodology. Structure (F5) was identified as the factor with the highest relative priority (overall weight=0.214), while adaptability (F4) (D-R=1.234) was found to be the most influencing aspect when selecting classifiers for technology adoption in patients with PD. In this case, the most appropriate algorithm for supporting technology adoption in patients with PD was the A3 - J48 decision tree (M3=2.5592). The results obtained by comparing the CoCoSo method in the proposed approach with 2 alternative methods (simple additive weighting and technique for order of preference by similarity to ideal solution) support the accuracy and applicability of the proposed methodology. It was observed that the final scores of the algorithms in each method were highly correlated (Pearson correlation coefficient >0.8).

Conclusions

The IF-AHP-IF-DEMATEL-CoCoSo approach helped to identify classification algorithms that do not just discriminate between good and bad adopters of assistive technologies within the Parkinson population but also consider technology-specific features like design, quality, and compatibility that make these classifiers easily implementable by clinicians in the health care system.

Predicting in-hospital mortality among patients admitted with a diagnosis of heart failure: a machine learning approach

Existing risk prediction models for hospitalized heart failure patients are limited. We identified patients hospitalized with a diagnosis of heart failure between 7 May 2013 and 26 April 2022 from a large academic, quaternary care medical centre (training cohort). Demographics, medical comorbidities, vitals, and labs were collected and were used to construct random forest machine learning models to predict in-hospital mortality. Models were compared with logistic regression, and to commonly used heart failure risk scores. The models were subsequently validated in patients hospitalized with a diagnosis of heart failure from a second academic, community medical centre (validation cohort). The entire cohort comprised 21 802 patients, of which 14 539 were in the training cohort and 7263 were in the validation cohort. The median age (25th-75th percentile) was 70 (58-82) for the entire cohort, 43.2% were female, and 6.7% experienced inpatient mortality. In the overall cohort, 7621 (35.0%) patients had heart failure with reduced ejection fraction (EF ≤ 40%), 1271 (5.8%) had heart failure with mildly reduced EF (EF 41-49%), and 12 910 (59.2%) had heart failure with preserved EF (EF ≥ 50%). Random forest models in the validation cohort demonstrated a c-statistic (95% confidence interval) of 0.96 (0.95-0.97), sensitivity (SN) of 87.3%, and specificity (SP) of 90.6% for the prediction of in-hospital mortality. Models for those with HFrEF demonstrated a c-statistic of 0.96 (0.94-0.98), SN 88.2%, and SP 91.0%, and those for patients with HFpEF showed a c-statistic of 0.95 (0.93-0.97), SN 87.4%, and SP 89.5% for predicting in-hospital mortality. The random forest model significantly outperformed logistic regression (c-statistic 0.87, SN 75.9%, and SP 86.9%), and current existing risk scores including the Acute Decompensated Heart Failure National Registry risk score (c-statistic of 0.70, SN 69%, and SP 62%), and the Get With the Guidelines-Heart Failure risk score (c-statistic 0.69, SN 67%, and SP 63%); P < 0.001 for comparison. Machine learning models built from commonly recorded patient information can accurately predict in-hospital mortality among patients hospitalized with a diagnosis of heart failure.

Integration of Patient Reported Quality-of-life Data into Risk Assessment in Heart Failure

BACKGROUND

Optimal management of outpatients with heart failure (HF) requires serially updating the estimates of their risk for adverse clinical outcomes to guide treatment. Patient-reported outcomes (PROs) are becoming increasingly used in clinical care. The purpose of this study was to determine whether the inclusion of PROs can improve the risk prediction for HF hospitalization and death in ambulatory patients with HF.

METHODS AND RESULTS

We included consecutive patients with HF with reduced ejection fraction (HFrEF) and HF with preserved EF (HFpEF) seen in a HF clinic between 2015 and 2019 who completed PROs as part of routine care. Cox regression with a least absolute shrinkage and selection operator regularization and gradient boosting machine analyses were used to estimate risk for a combined outcome of HF hospitalization, heart transplant, left ventricular assist device implantation, or death. The performance of the prediction models was evaluated with the time-dependent concordance index (Cτ). Among 1165 patients with HFrEF (mean age 59.1 ± 16.1, 68% male), the median follow-up was 487 days. Among 456 patients with HFpEF (mean age 64.2 ± 16.0 years, 55% male) the median follow-up was 494 days. Gradient boosting regression that included PROs had the best prediction performance - Cτ 0.73 for patients with HFrEF and 0.74 in patients with HFpEF, and showed very good stratification of risk by time to event analysis by quintile of risk. The Kansas City Cardiomyopathy Questionnaire overall summary score, visual analogue scale and Patient Reported Outcomes Measurement Information System dimensions of satisfaction with social roles and physical function had high variable importance measure in the models.

CONCLUSIONS

PROs improve risk prediction in both HFrEF and HFpEF, independent of traditional clinical factors. Routine assessment of PROs and leveraging the comprehensive data in the electronic health record in routine clinical care could help more accurately assess risk and support the intensification of treatment in patients with HF.

Performance of risk models to predict mortality risk for patients with heart failure: evaluation in an integrated health system

BACKGROUND

Referral of patients with heart failure (HF) who are at high mortality risk for specialist evaluation is recommended. Yet, most tools for identifying such patients are difficult to implement in electronic health record (EHR) systems.

OBJECTIVE

To assess the performance and ease of implementation of Machine learning Assessment of RisK and EaRly mortality in Heart Failure (MARKER-HF), a machine-learning model that uses structured data that is readily available in the EHR, and compare it with two commonly used risk scores: the Seattle Heart Failure Model (SHFM) and Meta-Analysis Global Group in Chronic (MAGGIC) Heart Failure Risk Score.

DESIGN

Retrospective, cohort study.

PARTICIPANTS

Data from 6764 adults with HF were abstracted from EHRs at a large integrated health system from 1/1/10 to 12/31/19.

MAIN MEASURES

One-year survival from time of first cardiology or primary care visit was estimated using MARKER-HF, SHFM, and MAGGIC. Discrimination was measured by the area under the receiver operating curve (AUC). Calibration was assessed graphically.

KEY RESULTS

Compared to MARKER-HF, both SHFM and MAGGIC required a considerably larger amount of data engineering and imputation to generate risk score estimates. MARKER-HF, SHFM, and MAGGIC exhibited similar discriminations with AUCs of 0.70 (0.69-0.73), 0.71 (0.69-0.72), and 0.71 (95% CI 0.70-0.73), respectively. All three scores showed good calibration across the full risk spectrum.

CONCLUSIONS

These findings suggest that MARKER-HF, which uses readily available clinical and lab measurements in the EHR and required less imputation and data engineering than SHFM and MAGGIC, is an easier tool to identify high-risk patients in ambulatory clinics who could benefit from referral to a HF specialist.

Admission blood glucose and 30-day mortality in patients with acute decompensated heart failure: prognostic significance in individuals with and without diabetes

Objective

Diabetes is a significant risk factor for acute heart failure, associated with an increased risk of mortality. This study aims to analyze the prognostic significance of admission blood glucose (ABG) on 30-day mortality in Chinese patients with acute decompensated heart failure (ADHF), with or without diabetes.

Methods

This retrospective study included 1,462 participants from the JX-ADHF1 cohort established between January 2019 to December 2022. We conducted multivariate cox regression, restricted cubic spline, receiver operating characteristic curve analysis, and mediation analysis to explore the association and potential mechanistic pathways (inflammation, oxidative stress, and nutrition) between ABG and 30-day mortality in ADHF patients, with and without diabetes.

Results

During the 30-day follow-up, we recorded 20 (5.36%) deaths in diabetic subjects and 33 (3.03%) in non-diabetics. Multivariate Cox regression revealed that ABG was independently associated with 30-day mortality in ADHF patients, with a stronger association in diabetics than non-diabetics (hazard ratio: Model 1: 1.71 vs 1.16; Model 2: 1.26 vs 1.19; Model 3: 1.65 vs 1.37; Model 4: 1.76 vs 1.33). Further restricted cubic spline analysis indicated a U-shaped relationship between ABG and 30-day mortality in non-diabetic ADHF patients (P for non-linearity < 0.001), with the lowest risk at ABG levels approximately between 5-7 mmol/L. Additionally, receiver operating characteristic analysis demonstrated that ABG had a higher predictive accuracy for 30-day mortality in diabetics (area under curve = 0.8751), with an optimal threshold of 13.95mmol/L. Finally, mediation analysis indicated a significant role of inflammation in ABG-related 30-day mortality in ADHF, accounting for 11.15% and 8.77% of the effect in diabetics and non-diabetics, respectively (P-value of proportion mediate < 0.05).

Conclusion

Our study confirms that ABG is a vital indicator for assessing and predicting 30-day mortality risk in ADHF patients with diabetes. For ADHF patients, both with and without diabetes, our evidence suggests that physicians should be alert and closely monitor any changes in patient conditions when ABG exceeds 13.95 mmol/L for those with diabetes and 7.05 mmol/L for those without. Timely adjustments in therapeutic strategies, including endocrine and anti-inflammatory treatments, are advisable.

Development and validation of a machine learning-based readmission risk prediction model for non-ST elevation myocardial infarction patients after percutaneous coronary intervention

To investigate the factors that influence readmissions in patients with acute non-ST elevation myocardial infarction (NSTEMI) after percutaneous coronary intervention (PCI) by using multiple machine learning (ML) methods to establish a predictive model. In this study, 1576 NSTEMI patients who were hospitalized at the Affiliated Hospital of North Sichuan Medical College were selected as the research subjects. They were divided into two groups: the readmitted group and the non-readmitted group. The division was based on whether the patients experienced complications or another incident of myocardial infarction within one year after undergoing PCI. Common variables selected by univariate and multivariate logistic regression, LASSO regression, and random forest were used as independent influencing factors for NSTEMI patients' readmissions after PCI. Six different ML models were constructed using these common variables. The area under the ROC curve, accuracy, sensitivity, and specificity were used to evaluate the performance of the six ML models. Finally, the optimal model was selected, and a nomogram was created to visually represent its clinical effectiveness. Three different methods were used to select seven representative common variables. These variables were then utilized to construct six different ML models, which were subsequently compared. The findings indicated that the LR model exhibited the most optimal performance in terms of AUC, accuracy, sensitivity, and specificity. The outcome, admission mode (walking and non-walking), communication ability, CRP, TC, HDL, and LDL were identified as independent predicators of readmissions in NSTEMI patients after PCI. The prediction model constructed by the LR algorithm was the best. The established column graph model established proved to be effective in identifying high-risk groups with high accuracy and differentiation. It holds a specific predictive value for the occurrence of readmissions after direct PCI in NSTEMI patients.

Improving the Prognostic Evaluation Precision of Hospital Outcomes for Heart Failure Using Admission Notes and Clinical Tabular Data: Multimodal Deep Learning Model

BACKGROUND

Clinical notes contain contextualized information beyond structured data related to patients' past and current health status.

OBJECTIVE

This study aimed to design a multimodal deep learning approach to improve the evaluation precision of hospital outcomes for heart failure (HF) using admission clinical notes and easily collected tabular data.

METHODS

Data for the development and validation of the multimodal model were retrospectively derived from 3 open-access US databases, including the Medical Information Mart for Intensive Care III v1.4 (MIMIC-III) and MIMIC-IV v1.0, collected from a teaching hospital from 2001 to 2019, and the eICU Collaborative Research Database v1.2, collected from 208 hospitals from 2014 to 2015. The study cohorts consisted of all patients with critical HF. The clinical notes, including chief complaint, history of present illness, physical examination, medical history, and admission medication, as well as clinical variables recorded in electronic health records, were analyzed. We developed a deep learning mortality prediction model for in-hospital patients, which underwent complete internal, prospective, and external evaluation. The Integrated Gradients and SHapley Additive exPlanations (SHAP) methods were used to analyze the importance of risk factors.

RESULTS

The study included 9989 (16.4%) patients in the development set, 2497 (14.1%) patients in the internal validation set, 1896 (18.3%) in the prospective validation set, and 7432 (15%) patients in the external validation set. The area under the receiver operating characteristic curve of the models was 0.838 (95% CI 0.827-0.851), 0.849 (95% CI 0.841-0.856), and 0.767 (95% CI 0.762-0.772), for the internal, prospective, and external validation sets, respectively. The area under the receiver operating characteristic curve of the multimodal model outperformed that of the unimodal models in all test sets, and tabular data contributed to higher discrimination. The medical history and physical examination were more useful than other factors in early assessments.

CONCLUSIONS

The multimodal deep learning model for combining admission notes and clinical tabular data showed promising efficacy as a potentially novel method in evaluating the risk of mortality in patients with HF, providing more accurate and timely decision support.

Predicting heart failure outcomes by integrating breath-by-breath measurements from cardiopulmonary exercise testing and clinical data through a deep learning survival neural network

Aims

Mathematical models previously developed to predict outcomes in patients with heart failure (HF) generally have limited performance and have yet to integrate complex data derived from cardiopulmonary exercise testing (CPET), including breath-by-breath data. We aimed to develop and validate a time-to-event prediction model using a deep learning framework using the DeepSurv algorithm to predict outcomes of HF.

Methods and results

Inception cohort of 2490 adult patients with high-risk cardiac conditions or HF underwent CPET with breath-by-breath measurements. Potential predictive features included known clinical indicators, standard summary statistics from CPETs, and mathematical features extracted from the breath-by-breath time series of 13 measurements. The primary outcome was a composite of death, heart transplant, or mechanical circulatory support treated as a time-to-event outcomes. Predictive features ranked as most important included many of the features engineered from the breath-by-breath data in addition to traditional clinical risk factors. The prediction model showed excellent performance in predicting the composite outcome with an area under the curve of 0.93 in the training and 0.87 in the validation data sets. Both the predicted vs. actual freedom from the composite outcome and the calibration of the prediction model were excellent. Model performance remained stable in multiple subgroups of patients.

Conclusion

Using a combined deep learning and survival algorithm, integrating breath-by-breath data from CPETs resulted in improved predictive accuracy for long-term (up to 10 years) outcomes in HF. DeepSurv opens the door for future prediction models that are both highly performing and can more fully use the large and complex quantity of data generated during the care of patients with HF.

The Efficacy of Machine Learning Models for Predicting the Prognosis of Heart Failure: A Systematic Review and Meta-Analysis

INTRODUCTION

Heart failure (HF) is a major global public health concern. The application of machine learning (ML) to identify individuals at high risk and enable early intervention is a promising approach for improving HF prognosis. We aim to systematically evaluate the performance and value of ML models for predicting HF prognosis.

METHODS

PubMed, Web of Science, Scopus, and Embase online databases were searched up to April 30, 2023, to identify studies on the use of ML models to predict HF prognosis. HF prognosis primarily encompasses readmission and mortality. The meta-analysis was conducted by MedCalc software. Subgroup analyses include grouping based on types of ML models, time intervals, sample sizes, the number of predictive variables, validation methods, whether to conduct hyperparameter optimization and calibration, data set partitioning methods.

RESULTS

A total of 31 studies were included. The most common ML models were random forest, boosting, support vector machine, neural network. The area under the receiver operating characteristic curve (AUC) for predicting HF readmission was 0.675 (95% CI: 0.651-0.699, p < 0.001), and the AUC for predicting HF mortality was 0.790 (95% CI: 0.765-0.816, p < 0.001). Subgroup analyses revealed that models with the prediction time interval of 1 year, sample sizes ≥10,000, the number of predictive variables ≥100, external validation, hyperparameter tuning, calibration adjustment, and data set partitioning using 10-fold cross-validation exhibited favorable performance within their respective subgroups.

CONCLUSION

The performance of ML models in predicting HF readmission is relatively poor, while its performance in predicting HF mortality is moderate. The quality of the relevant studies is generally low, it is essential to enhance the predictive capabilities of ML models through targeted improvements in practical applications.

The role of early-phase trials and real-world evidence in drug development

Phase 3 randomized controlled trials (RCTs), while the gold standard for treatment efficacy and safety, are not always feasible, are expensive, can be prolonged and can be limited in generalizability. Other under-recognized sources of evidence can also help advance drug development. Basic science, proof-of-concept studies and early-phase RCTs can provide evidence regarding the potential for clinical benefit. Real-world evidence generated from registries or observational datasets can provide insights into the treatment of rare diseases that often pose a challenge for trial recruitment. Pragmatic trials embedded in healthcare systems can assess the treatment effects in clinical settings among patient populations sometimes excluded from trials. This Perspective discusses potential sources of evidence that may be used to complement explanatory phase 3 RCTs and to speed the development of new cardiovascular medications. Content is derived from the 19th Global Cardiovascular Clinical Trialists meeting (December 2022), involving clinical trialists, patients, clinicians, regulators, funders and industry representatives.

Risk factors prediction of 6-month mortality after noncardiac surgery of older patients in China: a multicentre retrospective cohort study

BACKGROUND

Identifying the risk factors associated with perioperative mortality is crucial, particularly in older patients. Predicting 6-month mortality risk in older patients based on large datasets can assist patients and surgeons in perioperative clinical decision-making. This study aimed to develop a risk prediction model of mortality within 6 months after noncardiac surgery using the clinical data from 11 894 older patients in China.

MATERIALS AND METHODS

A multicentre, retrospective cohort study was conducted in 20 tertiary hospitals. The authors retrospectively included 11 894 patients (aged ≥65 years) who underwent noncardiac surgery between April 2020 and April 2022. The least absolute shrinkage and selection operator model based on linear regression was used to analyse and select risk factors, and various machine learning methods were used to build predictive models of 6-month mortality.

RESULTS

The authors predicted 12 preoperative risk factors associated with 6-month mortality in older patients after noncardiac surgery. Including laboratory-associated risk factors such as mononuclear cell ratio and total blood cholesterol level, etc. Also including medical history associated risk factors such as stroke, history of chronic diseases, etc. By using a random forest model, the authors constructed a predictive model with a satisfactory accuracy (area under the receiver operating characteristic curve=0.97).

CONCLUSION

The authors identified 12 preoperative risk factors associated with 6-month mortality in noncardiac surgery older patients. These preoperative risk factors may provide evidence for a comprehensive preoperative anaesthesia assessment as well as necessary information for clinical decision-making by anaesthesiologists.

Analysis of the Larissa Heart Failure Risk Score: Predictive Value in 9207 Patients Hospitalized for Heart Failure from a Single Center

Early risk stratification is of outmost clinical importance in hospitalized patients with heart failure (HHF). We examined the predictive value of the Larissa Heart Failure Risk Score (LHFRS) in a large population of HHF patients from the Cleveland Clinic. A total of 13,309 admissions for heart failure (HF) from 9207 unique patients were extracted from the Cleveland Clinic's electronic health record system. For each admission, components of the 3-variable simple LHFRS were obtained, including hypertension history, myocardial infarction history, and red blood cell distribution width (RDW) ≥ 15%. The primary outcome was a HF readmission and/or all-cause mortality at one year, and the secondary outcome was all-cause mortality at one year of discharge. For both outcomes, all variables were statistically significant, and the Kaplan-Meier curves were well-separated and in a consistent order (Log-rank test p-value < 0.001). Higher LHFRS values were found to be strongly related to patients experiencing an event, showing a clear association of LHFRS with this study outcomes. The bootstrapped-validated area under the curve (AUC) for the logistic regression model for each outcome revealed a C-index of 0.64 both for the primary and secondary outcomes, respectively. LHFRS is a simple risk model and can be utilized as a basis for risk stratification in patients hospitalized for HF.

Integration of machine learning to identify diagnostic genes in leukocytes for acute myocardial infarction patients

BACKGROUND

Acute myocardial infarction (AMI) has two clinical characteristics: high missed diagnosis and dysfunction of leukocytes. Transcriptional RNA on leukocytes is closely related to the course evolution of AMI patients. We hypothesized that transcriptional RNA in leukocytes might provide potential diagnostic value for AMI. Integration machine learning (IML) was first used to explore AMI discrimination genes. The following clinical study was performed to validate the results.

METHODS

A total of four AMI microarrays (derived from the Gene Expression Omnibus) were included in bioanalysis (220 sample size). Then, the clinical validation was finished with 20 AMI and 20 stable coronary artery disease patients (SCAD). At a ratio of 5:2, GSE59867 was included in the training set, while GSE60993, GSE62646, and GSE48060 were included in the testing set. IML was explicitly proposed in this research, which is composed of six machine learning algorithms, including support vector machine (SVM), neural network (NN), random forest (RF), gradient boosting machine (GBM), decision trees (DT), and least absolute shrinkage and selection operator (LASSO). IML had two functions in this research: filtered optimized variables and predicted the categorized value. Finally, The RNA of the recruited patients was analyzed to verify the results of IML.

RESULTS

Thirty-nine differentially expressed genes (DEGs) were identified between controls and AMI individuals from the training sets. Among the thirty-nine DEGs, IML was used to process the predicted classification model and identify potential candidate genes with overall normalized weights > 1. Finally, two genes (AQP9 and SOCS3) show their diagnosis value with the area under the curve (AUC) > 0.9 in both the training and testing sets. The clinical study verified the significance of AQP9 and SOCS3. Notably, more stenotic coronary arteries or severe Killip classification indicated higher levels of these two genes, especially SOCS3. These two genes correlated with two immune cell types, monocytes and neutrophils.

CONCLUSION

AQP9 and SOCS3 in leukocytes may be conducive to identifying AMI patients with SCAD patients. AQP9 and SOCS3 are closely associated with monocytes and neutrophils, which might contribute to advancing AMI diagnosis and shed light on novel genetic markers. Multiple clinical characteristics, multicenter, and large-sample relevant trials are still needed to confirm its clinical value.

Echocardiography-based machine learning algorithm for distinguishing ischemic cardiomyopathy from dilated cardiomyopathy

BACKGROUND

Machine learning (ML) can identify and integrate connections among data and has the potential to predict events. Heart failure is primarily caused by cardiomyopathy, and different etiologies require different treatments. The present study examined the diagnostic value of a ML algorithm that combines echocardiographic data to automatically differentiate ischemic cardiomyopathy (ICM) from dilated cardiomyopathy (DCM).

METHODS

We retrospectively collected the echocardiographic data of 200 DCM patients and 199 ICM patients treated in the First Affiliated Hospital of Guangxi Medical University between July 2016 and March 2022. All patients underwent invasive coronary angiography for diagnosis of ICM or DCM. The data were randomly divided into a training set and a test set via 10-fold cross-validation. Four ML algorithms (random forest, logistic regression, neural network, and XGBoost [ML algorithm under gradient boosting framework]) were used to generate a training model for the optimal subset, and the parameters were optimized. Finally, model performance was independently evaluated on the test set, and external validation was performed on 79 patients from another center.

RESULTS

Compared with the logistic regression model (area under the curve [AUC] = 0.925), neural network model (AUC = 0.893), and random forest model (AUC = 0.900), the XGBoost model had the best identification rate, with an average sensitivity of 72% and average specificity of 78%. The average accuracy was 75%, and the AUC of the optimal subset was 0.934. External validation produced an AUC of 0.804, accuracy of 78%, sensitivity of 64% and specificity of 93%.

CONCLUSIONS

We demonstrate that utilizing advanced ML algorithms can help to differentiate ICM from DCM and provide appreciable precision for etiological diagnosis and individualized treatment of heart failure patients.

Virtual healthcare solutions in heart failure: a literature review

The widespread adoption of mobile technologies offers an opportunity for a new approach to post-discharge care for patients with heart failure (HF). By enabling non-invasive remote monitoring and two-way, real-time communication between the clinic and home-based patients, as well as a host of other capabilities, mobile technologies have a potential to significantly improve remote patient care. This literature review summarizes clinical evidence related to virtual healthcare (VHC), defined as a care team + connected devices + a digital solution in post-release care of patients with HF. Searches were conducted on Embase (06/12/2020). A total of 171 studies were included for data extraction and evidence synthesis: 96 studies related to VHC efficacy, and 75 studies related to AI in HF. In addition, 15 publications were included from the search on studies scaling up VHC solutions in HF within the real-world setting. The most successful VHC interventions, as measured by the number of reported significant results, were those targeting reduction in rehospitalization rates. In terms of relative success rate, the two most effective interventions targeted patient self-care and all-cause hospital visits in their primary endpoint. Among the three categories of VHC identified in this review (telemonitoring, remote patient management, and patient self-empowerment) the integrated approach in remote patient management solutions performs the best in decreasing HF patients' re-admission rates and overall hospital visits. Given the increased amount of data generated by VHC technologies, artificial intelligence (AI) is being investigated as a tool to aid decision making in the context of primary diagnostics, identifying disease phenotypes, and predicting treatment outcomes. Currently, most AI algorithms are developed using data gathered in clinic and only a few studies deploy AI in the context of VHC. Most successes have been reported in predicting HF outcomes. Since the field of VHC in HF is relatively new and still in flux, this is not a typical systematic review capturing all published studies within this domain. Although the standard methodology for this type of reviews was followed, the nature of this review is qualitative. The main objective was to summarize the most promising results and identify potential research directions.

Artificial intelligence and heart failure: A state-of-the-art review

Heart failure (HF) is a heterogeneous syndrome affecting more than 60 million individuals globally. Despite recent advancements in understanding of the pathophysiology of HF, many issues remain including residual risk despite therapy, understanding the pathophysiology and phenotypes of patients with HF and preserved ejection fraction, and the challenges related to integrating a large amount of disparate information available for risk stratification and management of these patients. Risk prediction algorithms based on artificial intelligence (AI) may have superior predictive ability compared to traditional methods in certain instances. AI algorithms can play a pivotal role in the evolution of HF care by facilitating clinical decision making to overcome various challenges such as allocation of treatment to patients who are at highest risk or are more likely to benefit from therapies, prediction of adverse outcomes, and early identification of patients with subclinical disease or worsening HF. With the ability to integrate and synthesize large amounts of data with multidimensional interactions, AI algorithms can supply information with which physicians can improve their ability to make timely and better decisions. In this review, we provide an overview of the AI algorithms that have been developed for establishing early diagnosis of HF, phenotyping HF with preserved ejection fraction, and stratifying HF disease severity. This review also discusses the challenges in clinical deployment of AI algorithms in HF, and the potential path forward for developing future novel learning-based algorithms to improve HF care.

A Machine Learning Approach to Developing an Accurate Prediction of Maximal Heart Rate During Exercise Testing in Apparently Healthy Adults

PURPOSE

Maximal heart rate (HR max ) continues to be an important measure of adequate effort during an exercise test. The aim of this study was to improve the accuracy of HR max prediction using a machine learning (ML) approach.

METHODS

We used a sample from the Fitness Registry of the Importance of Exercise National Database, which included 17 325 apparently healthy individuals (81% males) who performed a maximal cardiopulmonary exercise test. Two standard formulas for HR max prediction were tested: Formula1 = 220 - age (yr), root-mean-squared error (RMSE) 21.9, relative root-mean-squared error (RRMSE) 1.1; and Formula2 = 209.3 - 0.72 × age (yr), RMSE 22.7 and RRMSE 1.1. For ML model prediction, we used age, weight, height, resting HR, and systolic and diastolic blood pressure. The following ML algorithms to predict HR max were applied: lasso regression (LR), neural networks (NN), support vector machine (SVM) and random forests (RF). An evaluation was performed using cross-validation and by computing the RMSE and RRMSE, Pearson correlation, and Bland-Altman plots. The best predictive model was explained with Shapley Additive Explanations (SHAP).

RESULTS

The HR max for the cohort was 162 ± 20 bpm. All ML models improved HR max prediction and reduced RMSE and RRMSE compared with Formula1 (LR: 20.2%, NN: 20.4%, SVM: 22.2%, and RF: 24.7%). The predictions of all algorithms significantly correlated with HR max ( r = 0.49, 0.51, 0.54, 0.57, respectively; P < .001). Bland-Altman analysis demonstrated lower bias and 95% CI for all ML models in comparison with standard equations. The SHAP explanation showed a high impact of all selected variables.

CONCLUSIONS

Machine learning, particularly the RF model, improved prediction of HR max using readily available measures. This approach should be considered for clinical application to refine HR max prediction.

Mortality Prediction in Patients With or Without Heart Failure Using a Machine Learning Model

Background

Most risk prediction models are confined to specific medical conditions, thus limiting their application to general medical populations.

Objectives

The MARKER-HF (Machine learning Assessment of RisK and EaRly mortality in Heart Failure) risk model was developed in heart failure (HF) patients. We assessed the ability of MARKER-HF to predict 1-year mortality in a large community-based hospital registry database including patients with and without HF.

Methods

This study included 41,749 consecutive patients who underwent echocardiography in a tertiary referral hospital (4,640 patients with and 37,109 without HF). Patients without HF were further subdivided into those with (n = 22,946) and without cardiovascular disease (n = 14,163) and also into cohorts based on recent acute coronary syndrome or history of atrial fibrillation, chronic obstructive pulmonary disease, chronic kidney disease, diabetes mellitus, hypertension, or malignancy.

Results

The median age of the 41,749 patients was 65 years, and 56.2% were male. The receiver operated area under the curves for MARKER-HF prediction of 1-year mortality of patients with HF was 0.729 (95% CI: 0.706-0.752) and for patients without HF was 0.770 (95% CI: 0.760-0.780). MARKER-HF prediction of mortality was consistent across subgroups with and without cardiovascular disease and in patients diagnosed with acute coronary syndrome, atrial fibrillation, chronic obstructive pulmonary disease, chronic kidney disease, diabetes mellitus, or hypertension. Patients with malignancy demonstrated higher mortality at a given MARKER-HF score than did patients in the other groups.

Conclusions

MARKER-HF predicts mortality for patients with HF as well as for patients suffering from a variety of diseases.

Emergency department risk assessment and disposition of acute heart failure patients: existing evidence and ongoing challenges

Heart failure (HF) is a global public health burden, characterized by frequent emergency department (ED) visits and hospitalizations. Identifying successful strategies to avoid admissions is crucial for the management of acutely decompensated HF, let alone resource utilization. The primary challenge for ED management of patients with acute heart failure (AHF) lies in the identification of those who can be safely discharged home instead of being admitted. This is an elaborate decision, based on limited objective evidence. Thus far, current biomarkers and risk stratification tools have had little impact on ED disposition decision-making. A reliable definition of a low-risk patient profile is warranted in order to accurately identify patients who could be appropriate for early discharge. A brief period of observation can facilitate risk stratification and allow for close monitoring, aggressive treatment, continuous assessment of response to initial therapy and patient education. Lung ultrasound may represent a valid bedside tool to monitor cardiogenic pulmonary oedema and determine the extent of achieved cardiac unloading after treatment in the observation unit setting. Safe discharge mandates multidisciplinary collaboration and thoughtful assessment of socioeconomic and behavioural factors, along with a clear post-discharge plan put forward and a close follow-up in an outpatient setting. Ongoing research to improve ED risk stratification and disposition of AHF patients may mitigate the tremendous public health challenge imposed by the HF epidemic.

Machine Learning for Mortality Prediction in Patients With Heart Failure With Mildly Reduced Ejection Fraction

Background Machine-learning-based prediction models (MLBPMs) have shown satisfactory performance in predicting clinical outcomes in patients with heart failure with reduced and preserved ejection fraction. However, their usefulness has yet to be fully elucidated in patients with heart failure with mildly reduced ejection fraction. This pilot study aims to evaluate the prediction performance of MLBPMs in a heart failure with mildly reduced ejection fraction cohort with long-term follow-up data. Methods and Results A total of 424 patients with heart failure with mildly reduced ejection fraction were enrolled in our study. The primary outcome was all-cause mortality. Two feature selection strategies were introduced for MLBPM development. The "All-in" (67 features) strategy was based on feature correlation, multicollinearity, and clinical significance. The other strategy was the CoxBoost algorithm with 10-fold cross-validation (17 features), which was based on the selection result of the "All-in" strategy. Six MLBPMs with 5-fold cross-validation based on the "All-in" and the CoxBoost algorithm with 10-fold cross-validation strategy were developed by the eXtreme Gradient Boosting, random forest, and support vector machine algorithms. The logistic regression model with 14 benchmark predictors was used as a reference model. During a median follow-up of 1008 (750, 1937) days, 121 patients met the primary outcome. Overall, MLBPMs outperformed the logistic model. The "All-in" eXtreme Gradient Boosting model had the best performance, with an accuracy of 85.4% and a precision of 70.3%. The area under the receiver-operating characteristic curve was 0.916 (95% CI, 0.887-0.945). The Brier score was 0.12. Conclusions The MLBPMs could significantly improve outcome prediction in patients with heart failure with mildly reduced ejection fraction, which would further optimize the management of these patients.

What drives performance in machine learning models for predicting heart failure outcome?

Aims

The development of acute heart failure (AHF) is a critical decision point in the natural history of the disease and carries a dismal prognosis. The lack of appropriate risk-stratification tools at hospital discharge of AHF patients significantly limits clinical ability to precisely tailor patient-specific therapeutic regimen at this pivotal juncture. Machine learning-based strategies may improve risk stratification by incorporating analysis of high-dimensional patient data with multiple covariates and novel prediction methodologies. In the current study, we aimed at evaluating the drivers for success in prediction models and establishing an institute-tailored artificial Intelligence-based prediction model for real-time decision support.

Methods and results

We used a cohort of all 10 868 patients AHF patients admitted to a tertiary hospital during a 12 years period. A total of 372 covariates were collected from admission to the end of the hospitalization. We assessed model performance across two axes: (i) type of prediction method and (ii) type and number of covariates. The primary outcome was 1-year survival from hospital discharge. For the model-type axis, we experimented with seven different methods: logistic regression (LR) with either L1 or L2 regularization, random forest (RF), Cox proportional hazards model (Cox), extreme gradient boosting (XGBoost), a deep neural-net (NeuralNet) and an ensemble classifier of all the above methods. We were able to achieve an area under receiver operator curve (AUROC) prediction accuracy of more than 80% with most prediction models including L1/L2-LR (80.4%/80.3%), Cox (80.2%), XGBoost (80.5%), NeuralNet (80.4%). RF was inferior to other methods (78.8%), and the ensemble model was slightly superior (81.2%). The number of covariates was a significant modifier (P < 0.001) of prediction success, the use of multiplex-covariates preformed significantly better (AUROC 80.4% for L1-LR) compared with a set of known clinical covariates (AUROC 77.8%). Demographics followed by lab-tests and administrative data resulted in the largest gain in model performance.

Conclusions

The choice of the predictive modelling method is secondary to the multiplicity and type of covariates for predicting AHF prognosis. The application of a structured data pre-processing combined with the use of multiple-covariates results in an accurate, institute-tailored risk prediction in AHF.

Self-report symptom-based endometriosis prediction using machine learning

Endometriosis is a chronic gynecological condition that affects 5-10% of reproductive age women. Nonetheless, the average time-to-diagnosis is usually between 6 and 10 years from the onset of symptoms. To shorten time-to-diagnosis, many studies have developed non-invasive screening tools. However, most of these studies have focused on data obtained from women who had/were planned for laparoscopy surgery, that is, women who were near the end of the diagnostic process. In contrast, our study aimed to develop a self-diagnostic tool that predicts the likelihood of endometriosis based only on experienced symptoms, which can be used in early stages of symptom onset. We applied machine learning to train endometriosis prediction models on data obtained via questionnaires from two groups of women: women who were diagnosed with endometriosis and women who were not diagnosed. The best performing model had AUC of 0.94, sensitivity of 0.93, and specificity of 0.95. The model is intended to be incorporated into a website as a self-diagnostic tool and is expected to shorten time-to-diagnosis by referring women with a high likelihood of having endometriosis to further examination. We also report the importance and effectiveness of different symptoms in predicting endometriosis.

A hybrid ensemble approach to accelerate the classification accuracy for predicting malnutrition among under-five children in sub-Saharan African countries

BACKGROUND

The proper intake of nutrients is essential to the growth and maturation of youngsters. In sub-Saharan Africa, 1 in 7 children dies before age 5 y, and more than a third of these deaths are attributed to malnutrition. The main purpose of this study was to develop a majority voting-based hybrid ensemble (MVBHE) learning model to accelerate the prediction accuracy of malnutrition data of under-five children in sub-Saharan Africa.

METHODS

This study used available under-five nutritional secondary data from the Demographic and Health Surveys performed in sub-Saharan African countries. The research used bagging, boosting, and voting algorithms, such as random forest, decision tree, eXtreme Gradient Boosting, and k-nearest neighbors machine learning methods, to generate the MVBHE model.

RESULTS

We evaluated the model performances in contrast to each other using different measures, including accuracy, precision, recall, and the F1 score. The results of the experiment showed that the MVBHE model (96%) was better at predicting malnutrition than the random forest (81%), decision tree (60%), eXtreme Gradient Boosting (79%), and k-nearest neighbors (74%).

CONCLUSIONS

The random forest algorithm demonstrated the highest prediction accuracy (81%) compared with the decision tree, eXtreme Gradient Boosting, and k-nearest neighbors algorithms. The accuracy was then enhanced to 96% using the MVBHE model. The MVBHE model is recommended by the present study as the best way to predict malnutrition in under-five children.

Predicting heart failure in-hospital mortality by integrating longitudinal and category data in electronic health records

Heart failure is a life-threatening syndrome that is diagnosed in 3.6 million people worldwide each year. We propose a deep fusion learning model (DFL-IMP) that uses time series and category data from electronic health records to predict in-hospital mortality in patients with heart failure. We considered 41 time series features (platelets, white blood cells, urea nitrogen, etc.) and 17 category features (gender, insurance, marital status, etc.) as predictors, all of which were available within the time of the patient's last hospitalization, and a total of 7696 patients participated in the observational study. Our model was evaluated against different time windows. The best performance was achieved with an AUC of 0.914 when the observation window was 5 days and the prediction window was 30 days. Outperformed other baseline models including LR (0.708), RF (0.717), SVM (0.675), LSTM (0.757), GRU (0.759), GRU-U (0.766) and MTSSP (0.770). This tool allows us to predict the expected pathway of heart failure patients and intervene early in the treatment process, which has significant implications for improving the life expectancy of heart failure patients.

A deep learning system for heart failure mortality prediction

Heart failure (HF) is the final stage of the various heart diseases developing. The mortality rates of prognosis HF patients are highly variable, ranging from 5% to 75%. Evaluating the all-cause mortality of HF patients is an important means to avoid death and positively affect the health of patients. But in fact, machine learning models are difficult to gain good results on missing values, high dimensions, and imbalances HF data. Therefore, a deep learning system is proposed. In this system, we propose an indicator vector to indicate whether the value is true or be padded, which fast solves the missing values and helps expand data dimensions. Then, we use a convolutional neural network with different kernel sizes to obtain the features information. And a multi-head self-attention mechanism is applied to gain whole channel information, which is essential for the system to improve performance. Besides, the focal loss function is introduced to deal with the imbalanced problem better. The experimental data of the system are from the public database MIMIC-III, containing valid data for 10311 patients. The proposed system effectively and fast predicts four death types: death within 30 days, death within 180 days, death within 365 days and death after 365 days. Our study uses Deep SHAP to interpret the deep learning model and obtains the top 15 characteristics. These characteristics further confirm the effectiveness and rationality of the system and help provide a better medical service.

Clinical application of artificial intelligence algorithm for prediction of one-year mortality in heart failure patients

Risk prediction for heart failure (HF) using machine learning methods (MLM) has not yet been established at practical application levels in clinical settings. This study aimed to create a new risk prediction model for HF with a minimum number of predictor variables using MLM. We used two datasets of hospitalized HF patients: retrospective data for creating the model and prospectively registered data for model validation. Critical clinical events (CCEs) were defined as death or LV assist device implantation within 1 year from the discharge date. We randomly divided the retrospective data into training and testing datasets and created a risk prediction model based on the training dataset (MLM-risk model). The prediction model was validated using both the testing dataset and the prospectively registered data. Finally, we compared predictive power with published conventional risk models. In the patients with HF (n = 987), CCEs occurred in 142 patients. In the testing dataset, the substantial predictive power of the MLM-risk model was obtained (AUC = 0.87). We generated the model using 15 variables. Our MLM-risk model showed superior predictive power in the prospective study compared to conventional risk models such as the Seattle Heart Failure Model (c-statistics: 0.86 vs. 0.68, p < 0.05). Notably, the model with an input variable number (n = 5) has comparable predictive power for CCE with the model (variable number = 15). This study developed and validated a model with minimized variables to predict mortality more accurately in patients with HF, using a MLM, than the existing risk scores.

Predicting Six-Month Re-Admission Risk in Heart Failure Patients Using Multiple Machine Learning Methods: A Study Based on the Chinese Heart Failure Population Database

Since most patients with heart failure are re-admitted to the hospital, accurately identifying the risk of re-admission of patients with heart failure is important for clinical decision making and management. This study plans to develop an interpretable predictive model based on a Chinese population for predicting six-month re-admission rates in heart failure patients. Research data were obtained from the PhysioNet portal. To ensure robustness, we used three approaches for variable selection. Six different machine learning models were estimated based on selected variables. The ROC curve, prediction accuracy, sensitivity, and specificity were used to evaluate the performance of the established models. In addition, we visualized the optimized model with a nomogram. In all, 2002 patients with heart failure were included in this study. Of these, 773 patients experienced re-admission and a six-month re-admission incidence of 38.61%. Based on evaluation metrics, the logistic regression model performed best in the validation cohort, with an AUC of 0.634 (95%CI: 0.599-0.646) and an accuracy of 0.652. A nomogram was also generated. The established prediction model has good discrimination ability in predicting. Our findings are helpful and could provide useful information for the allocation of healthcare resources and for improving the quality of survival of heart failure patients.

Stroke prevention in rural residents: development of a simplified risk assessment tool with artificial intelligence

BACKGROUND

Limited studies have focused on the risk assessment of stroke in rural regions. Moreover, the application of artificial intelligence in stroke risk scoring system is still insufficient. This study aims to develop a simplified and visualized risk score with good performance and convenience for rural stroke risk assessment, which is combined with a machine learning (ML) algorithm.

METHODS

Participants of the Henan Rural Cohort were enrolled in this study. The total participants (n = 38,322) were randomly split into a train set and a test set in the ratio of 7:3. An ML algorithm was used to select variables and the logistic regression was then applied to construct the scoring system. The C-statistic and the Brier score (BS) were used to evaluate the discrimination and calibration. The Framingham stroke risk profile (FSRP) and the self-reported stroke risk function (SRSRF) were chosen to be compared.

RESULTS

The Rural Stroke Risk Score (RSRS) was produced in this study, including age, drinking status, triglyceride, type 2 diabetes mellitus, hypertension, waist circumference, and family history of stroke. On validation, the C-statistic was 0.757 (95% CI 0.749-0.765) and the BS was 0.058 in the test set. In addition, the discrimination of RSRS was 6.02% and 7.34% higher than that of the FSRP and SRSRF, respectively.

CONCLUSIONS

A well-performed scoring system for assessing stroke risk in rural residents was developed in this study. This risk score would facilitate stroke screening and the prevention of cardiovascular disease in economically underdeveloped areas.

Predicting in-hospital all-cause mortality in heart failure using machine learning

Background

The age of onset and causes of heart failure differ between high-income and low-and-middle-income countries (LMIC). Heart failure patients in LMIC also experience a higher mortality rate. Innovative ways that can risk stratify heart failure patients in this region are needed. The aim of this study was to demonstrate the utility of machine learning in predicting all-cause mortality in heart failure patients hospitalised in a tertiary academic centre.

Methods

Six supervised machine learning algorithms were trained to predict in-hospital all-cause mortality using data from 500 consecutive heart failure patients with a left ventricular ejection fraction (LVEF) less than 50%.

Results

The mean age was 55.2 ± 16.8 years. There were 271 (54.2%) males, and the mean LVEF was 29 ± 9.2%. The median duration of hospitalisation was 7 days (interquartile range: 4-11), and it did not differ between patients discharged alive and those who died. After a prediction window of 4 years (interquartile range: 2-6), 84 (16.8%) patients died before discharge from the hospital. The area under the receiver operating characteristic curve was 0.82, 0.78, 0.77, 0.76, 0.75, and 0.62 for random forest, logistic regression, support vector machines (SVM), extreme gradient boosting, multilayer perceptron (MLP), and decision trees, and the accuracy during the test phase was 88, 87, 86, 82, 78, and 76% for random forest, MLP, SVM, extreme gradient boosting, decision trees, and logistic regression. The support vector machines were the best performing algorithm, and furosemide, beta-blockers, spironolactone, early diastolic murmur, and a parasternal heave had a positive coefficient with the target feature, whereas coronary artery disease, potassium, oedema grade, ischaemic cardiomyopathy, and right bundle branch block on electrocardiogram had negative coefficients.

Conclusion

Despite a small sample size, supervised machine learning algorithms successfully predicted all-cause mortality with modest accuracy. The SVM model will be externally validated using data from multiple cardiology centres in South Africa before developing a uniquely African risk prediction tool that can potentially transform heart failure management through precision medicine.

Heart failure: an update from the last years and a look at the near future

In the last years, major progress occurred in heart failure (HF) management. Quadruple therapy is now mandatory for all the patients with HF with reduced ejection fraction. Whilst verciguat is becoming available across several countries, omecamtiv mecarbil is waiting to be released for clinical use. Concurrent use of potassium-lowering agents may counteract hyperkalaemia and facilitate renin-angiotensin-aldosterone system inhibitor implementations. The results of the EMPagliflozin outcomE tRial in Patients With chrOnic heaRt Failure With Preserved Ejection Fraction (EMPEROR-Preserved) trial were confirmed by the Dapagliflozin in Heart Failure with Mildly Reduced or Preserved Ejection Fraction (DELIVER) trial, and we now have, for the first time, evidence for treatment of also patients with HF with preserved ejection fraction. In a pre-specified meta-analysis of major randomized controlled trials, sodium-glucose co-transporter-2 inhibitors reduced all-cause mortality, cardiovascular (CV) mortality, and HF hospitalization in the patients with HF regardless of left ventricular ejection fraction. Other steps forward have occurred in the treatment of decompensated HF. Acetazolamide in Acute Decompensated Heart Failure with Volume Overload (ADVOR) trial showed that the addition of intravenous acetazolamide to loop diuretics leads to greater decongestion vs. placebo. The addition of hydrochlorothiazide to loop diuretics was evaluated in the CLOROTIC trial. Torasemide did not change outcomes, compared with furosemide, in TRANSFORM-HF. Ferric derisomaltose had an effect on the primary outcome of CV mortality or HF rehospitalizations in IRONMAN (rate ratio 0.82; 95% confidence interval 0.66-1.02; P = 0.070). Further options for the treatment of HF, including device therapies, cardiac contractility modulation, and percutaneous treatment of valvulopathies, are summarized in this article.

Risk Association of Liver Cancer and Hepatitis B with Tree Ensemble and Lifestyle Features

The second-largest cause of death by cancer in Korea is liver cancer, which leads to acute morbidity and mortality. Hepatitis B is the most common cause of liver cancer. About 70% of liver cancer patients suffer from hepatitis B. Early risk association of liver cancer and hepatitis B can help prevent fatal conditions. We propose a risk association method for liver cancer and hepatitis B with only lifestyle features. The diagnostic features were excluded to reduce the cost of gathering medical data. The data source is the Korea National Health and Nutrition Examination Survey (KNHANES) from 2007 to 2019. We use 3872 and 4640 subjects for liver cancer and hepatitis B model, respectively. Random forest is employed to determine functional relationships between liver diseases and lifestyle features. The performance of our proposed method was compared with six machine learning methods. The results showed the proposed method outperformed the other methods in the area under the receiver operator characteristic curve of 0.8367. The promising results confirm the superior performance of the proposed method and show that the proposed method with only lifestyle features provides significant advantages, potentially reducing the cost of detecting patients who require liver health care in advance.

Performance of current risk stratification models for predicting mortality in patients with heart failure: a systematic review and meta-analysis

AIMS

There are several risk scores designed to predict mortality in patients with heart failure (HF). This study aimed to assess performance of risk scores validated for mortality prediction in patients with acute HF (AHF) and chronic HF.

METHODS AND RESULTS

MEDLINE and Scopus were searched from January 2015 to January 2021 for studies which internally or externally validated risk models for predicting all-cause mortality in patients with AHF and chronic HF. Discrimination data were analysed using C-statistics, and pooled using generic inverse-variance random-effects model. Nineteen studies (n = 494 156 patients; AHF: 24 762; chronic HF mid-term mortality: 62 000; chronic HF long-term mortality: 452 097) and 11 risk scores were included. Overall, discrimination of risk scores was good across the three subgroups: AHF mortality [C-statistic: 0.76 (0.68-0.83)], chronic HF mid-term mortality [1 year; C-statistic: 0.74 (0.68-0.79)], and chronic HF long-term mortality [≥2 years; C-statistic: 0.71 (0.69-0.73)]. MEESSI-AHF [C-statistic: 0.81 (0.80-0.83)] and MARKER-HF [C-statistic: 0.85 (0.80-0.89)] had an excellent discrimination for AHF and chronic HF mid-term mortality, respectively, whereas MECKI had good discrimination [C-statistic: 0.78 (0.73-0.83)] for chronic HF long-term mortality relative to other models. Overall, risk scores predicting short-term mortality in patients with AHF did not have evidence of poor calibration (Hosmer-Lemeshow P > 0.05). However, risk models predicting mid-term and long-term mortality in patients with chronic HF varied in calibration performance.

CONCLUSIONS

The majority of recently validated risk scores showed good discrimination for mortality in patients with HF. MEESSI-AHF demonstrated excellent discrimination in patients with AHF, and MARKER-HF and MECKI displayed an excellent discrimination in patients with chronic HF. However, modest reporting of calibration and lack of head-to-head comparisons in same populations warrant future studies.

Augmented Intelligence to Identify Patients With Advanced Heart Failure in an Integrated Health System

BACKGROUND

Timely referral for specialist evaluation in patients with advanced heart failure (HF) is a Class 1 recommendation. However, the transition from stage C HF to advanced or stage D HF often goes undetected in routine care, resulting in delayed referral and higher mortality rates.

OBJECTIVES

The authors sought to develop an augmented intelligence-enabled workflow using machine learning to identify patients with stage D HF and streamline referral.

METHODS

We extracted data on HF patients with encounters from January 1, 2007, to November 30, 2020, from a HF registry within a regional, integrated health system. We created an ensemble machine learning model to predict stage C or stage D HF and integrated the results within the electronic health record.

RESULTS

In a retrospective data set of 14,846 patients, the model had a good positive predictive value (60%) and low sensitivity (25%) for identifying stage D HF in a 100-person, physician-reviewed, holdout test set. During prospective implementation of the workflow from April 1, 2021, to February 15, 2022, 416 patients were reviewed by a clinical coordinator, with agreement between the model and the coordinator in 50.3% of stage D predictions. Twenty-four patients have been scheduled for evaluation in a HF clinic, 4 patients started an evaluation for advanced therapies, and 1 patient received a left ventricular assist device.

CONCLUSIONS

An augmented intelligence-enabled workflow was integrated into clinical operations to identify patients with advanced HF. Endeavors such as this require a multidisciplinary team with experience in design thinking, informatics, quality improvement, operations, and health information technology, as well as dedicated resources to monitor and improve performance over time.

Machine learning-derived prediction of in-hospital mortality in patients with severe acute respiratory infection: analysis of claims data from the German-wide Helios hospital network

BACKGROUND

Severe acute respiratory infections (SARI) are the most common infectious causes of death. Previous work regarding mortality prediction models for SARI using machine learning (ML) algorithms that can be useful for both individual risk stratification and quality of care assessment is scarce. We aimed to develop reliable models for mortality prediction in SARI patients utilizing ML algorithms and compare its performances with a classic regression analysis approach.

METHODS

Administrative data (dataset randomly split 75%/25% for model training/testing) from years 2016-2019 of 86 German Helios hospitals was retrospectively analyzed. Inpatient SARI cases were defined by ICD-codes J09-J22. Three ML algorithms were evaluated and its performance compared to generalized linear models (GLM) by computing receiver operating characteristic area under the curve (AUC) and area under the precision-recall curve (AUPRC).

RESULTS

The dataset contained 241,988 inpatient SARI cases (75 years or older: 49%; male 56.2%). In-hospital mortality was 11.6%. AUC and AUPRC in the testing dataset were 0.83 and 0.372 for GLM, 0.831 and 0.384 for random forest (RF), 0.834 and 0.382 for single layer neural network (NNET) and 0.834 and 0.389 for extreme gradient boosting (XGBoost). Statistical comparison of ROC AUCs revealed a better performance of NNET and XGBoost as compared to GLM.

CONCLUSION

ML algorithms for predicting in-hospital mortality were trained and tested on a large real-world administrative dataset of SARI patients and showed good discriminatory performances. Broad application of our models in clinical routine practice can contribute to patients' risk assessment and quality management.

An Artificial Intelligence Approach to Guiding the Management of Heart Failure Patients Using Predictive Models: A Systematic Review

Heart failure (HF) is one of the leading causes of mortality and hospitalization worldwide. The accurate prediction of mortality and readmission risk provides crucial information for guiding decision making. Unfortunately, traditional predictive models reached modest accuracy in HF populations. We therefore aimed to present predictive models based on machine learning (ML) techniques in HF patients that were externally validated. We searched four databases and the reference lists of the included papers to identify studies in which HF patient data were used to create a predictive model. Literature screening was conducted in Academic Search Ultimate, ERIC, Health Source Nursing/Academic Edition and MEDLINE. The protocol of the current systematic review was registered in the PROSPERO database with the registration number CRD42022344855. We considered all types of outcomes: mortality, rehospitalization, response to treatment and medication adherence. The area under the receiver operating characteristic curve (AUC) was used as the comparator parameter. The literature search yielded 1649 studies, of which 9 were included in the final analysis. The AUCs for the machine learning models ranged from 0.6494 to 0.913 in independent datasets, whereas the AUCs for statistical predictive scores ranged from 0.622 to 0.806. Our study showed an increasing number of ML predictive models concerning HF populations, although external validation remains infrequent. However, our findings revealed that ML approaches can outperform conventional risk scores and may play important role in HF management.