scholarly journals Deep learning for mortality prediction in patients with a de-novo or worsened heart failure

2021 ◽  
Vol 42 (Supplement_1) ◽  
Author(s):  
R Herman ◽  
M Vanderheyden ◽  
B Vavrik ◽  
M Beles ◽  
T Palus ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Heart Failure ◽  
De Novo ◽  
Care Pathway ◽  
Mortality Prediction ◽  
Risk Scores ◽  
Stable Model ◽  
Patient Level ◽  
Clinical Risk ◽  
All Cause Mortality

Abstract Background Heart failure (HF) is a heterogenous syndrome with complex pathophysiology. Biomarkers and clinical risk scores often fail to provide optimal patient-level precision in the prognostic stratification. As utilizing single observational timepoint, they do not capture the entire care pathway with variations in individual patient management. Electronic patient records provide an opportunity to develop new artificial intelligence (AI) strategies for comprehensive prognostic re-stratification reflecting diagnostic and therapeutic management. Purpose We sought to use deep artificial intelligence (AI) and develop an unbiased predictive algorithm for all-cause mortality in a cohort of patients hospitalized with a de novo or worsened HF. Methods In a cohort of 2449 HF patients hospitalized between 2011–2017, we utilized 151 451 patient exams from 422 parameters. They included clinical phenotyping, medication, ECG, laboratory, echocardiography, catheterization data or percutaneous and surgical interventions gathered on a routine clinical basis reflecting standard of care as captured in individual electronic records. The AI model development consisted of 101 iterations of repeated random subsampling splits into balanced training and validation sets. Results AI models yielded performance ranging from 0.83 to 0.89 AUC on the outcome-balanced validation set in predicting all-cause mortality at 30-, 90-, 180-, 360- and 720-day time-limits (Figure 1). The primary endpoint, 1-year mortality prediction model, recorded an 0.85 AUC accuracy. We observed stable model performance across all HF phenotypes: HFpEF 0.83 AUC, HFmrEF 0.85 AUC and HFrEF 0.86 AUC, respectively). Conclusion Our findings present a novel, patient-level, AI-based risk prediction of all-cause mortality in heart failure with a robust accuracy across its phenotypes. This suggests the potential of AI based predictive models in a point-of-care approach to guide clinical risk stratification. FUNDunding Acknowledgement Type of funding sources: Foundation. Main funding source(s): VZW Cardiovascular Research Center Aalst

Incorporation of Natriuretic Peptides with Clinical Risk‐scores to Predict Heart Failure Among Individuals with Dysglycemia

2021 ◽  
Author(s):  
Matthew W. Segar ◽  
Muhammad Shahzeb Khan ◽  
Kershaw V. Patel ◽  
Muthiah Vaduganathan ◽  
Vaishnavi Kannan ◽  
...  
Keyword(s):  
Heart Failure ◽  
Natriuretic Peptides ◽  
Risk Scores ◽  
Clinical Risk

Abstract 14541: Incidence, Risk Factors, and Outcomes of Heart Failure in Patients With Graves’ Disease

Circulation ◽  
2020 ◽  
Vol 142 (Suppl_3) ◽  
Author(s):  
Jwan A Naser ◽  
Sorin Pislaru ◽  
Marius N Stan ◽  
Grace Lin
Keyword(s):  
Risk Factors ◽  
Heart Failure ◽  
Cardiovascular Events ◽  
De Novo ◽  
Left Ventricular ◽  
Left Ventricular Ejection ◽  
Graves Disease ◽  
Increased Risk ◽  
All Cause Mortality ◽  
Overt Hyperthyroidism

Background: Graves’ disease (GD) can both aggravate pre-existing cardiac disease and cause de novo heart failure (HF). Due to the rarity of thyrotoxic HF, population-based studies are lacking, and data from smaller studies are widely variable. Methods: We reviewed the medical records of 1371 consecutive patients with GD evaluated at our clinic between 2009 and 2019. HF was defined according to Framingham criteria. GD-related HFrEF was defined by left ventricular ejection fraction of <50%, while HFpEF was defined according to the Heart Failure Association of the European Society of Cardiology. Outcomes of major cardiovascular events, all-cause mortality, and cardiac hospitalizations were analyzed with adjustments for age, gender, and history of coronary artery disease (CAD). 1:1 matching with controls (age, gender, and CAD) was additionally done. Results: HF occurred in 74 patients (31 HFrEF; 43 HFpEF). Incidence of GD-related HF, HFrEF, and HFpEF was 5.4%, 2.3%, and 3.1%, respectively. In HFrEF, atrial fibrillation (AF) (RR 10.05, p <0.001) and thyrotropin receptor antibodies (TRAb) level (RR 1.05 per unit, p=0.005) were independent predisposing factors. In HFpEF, independent risk factors were COPD (RR 5.78, p < 0.001), older age (RR 1.48 per 10 years, p = 0.003), overt hyperthyroidism (RR 5.37, p = 0.021), higher BMI (1.06 per unit, p = 0.003), and HTN (RR 3.03, p = 0.011). Rates of cardiac hospitalizations were higher in HFrEF (41.9% vs 3.2%, p <0.001) and HFpEF (44.2% vs 4.7%, p < 0.001) compared to controls. Furthermore, while both increased risk of strokes (HFrEF: RR 4.12, p = 0.027; HFpEF: RR 4.64, p = 0.009), only HFrEF increased risk of all-cause mortality (RR 3.78, p = 0.045). Conclusion: De novo HF occurs in 5.4% of patients with GD and increases the rate of cardiovascular events. HF occurs more frequently in GD patients with AF, higher TRAb, higher BMI, and overt hyperthyroidism, suggesting that these may be targets for treatment to prevent cardiovascular complications, especially in older multimorbid patients.

scholarly journals Improving risk stratification in heart failure with preserved ejection fraction by combining two validated risk scores

Open Heart ◽  
2019 ◽  
Vol 6 (1) ◽  
pp. e000961 ◽  
Author(s):  
Kalyani Anil Boralkar ◽  
Yukari Kobayashi ◽  
Kegan J Moneghetti ◽  
Vedant S Pargaonkar ◽  
Mirela Tuzovic ◽  
...  
Keyword(s):  
Heart Failure ◽  
Ejection Fraction ◽  
Risk Score ◽  
Multivariable Analysis ◽  
Laboratory Data ◽  
Primary Diagnosis ◽  
Risk Scores ◽  
Preserved Ejection Fraction ◽  
Hospitalised Patients ◽  
All Cause Mortality

IntroductionThe Intermountain Risk Score (IMRS) was developed and validated to predict short-term and long-term mortality in hospitalised patients using demographics and commonly available laboratory data. In this study, we sought to determine whether the IMRS also predicts all-cause mortality in patients hospitalised with heart failure with preserved ejection fraction (HFpEF) and whether it is complementary to the Get with the Guidelines Heart Failure (GWTG-HF) risk score or N-terminal pro-B-type natriuretic peptide (NT-proBNP).Methods and resultsWe used the Stanford Translational Research Integrated Database Environment to identify 3847 adult patients with a diagnosis of HFpEF between January 1998 and December 2016. Of these, 580 were hospitalised with a primary diagnosis of acute HFpEF. Mean age was 76±16 years, the majority being female (58%), with a high prevalence of diabetes mellitus (36%) and a history of coronary artery disease (60%). Over a median follow-up of 2.0 years, 140 (24%) patients died. On multivariable analysis, the IMRS and GWTG-HF risk score were independently associated with all-cause mortality (standardised HRs IMRS (1.55 (95% CI 1.27 to 1.93)); GWTG-HF (1.60 (95% CI 1.27 to 2.01))). Combining the two scores, improved the net reclassification over GWTG-HF alone by 36.2%. In patients with available NT-proBNP (n=341), NT-proBNP improved the net reclassification of each score by 46.2% (IMRS) and 36.3% (GWTG-HF).ConclusionIMRS and GWTG-HF risk scores, along with NT-proBNP, play a complementary role in predicting outcome in patients hospitalised with HFpEF.

scholarly journals Hemodynamic profiles by non-invasive monitoring of cardiac index and vascular tone in acute heart failure patients in the emergency department: external validation and clinical outcomes

2021 ◽  
Author(s):  
Nicholas Eric Harrison ◽  
Sarah Meram ◽  
Xiangrui Li ◽  
Patrick Medado ◽  
Morgan B White ◽  
...  
Keyword(s):  
Heart Failure ◽  
Emergency Department ◽  
Cardiac Index ◽  
Acute Heart Failure ◽  
Vascular Tone ◽  
Primary Outcome ◽  
De Novo ◽  
Risk Scores ◽  
Secondary Outcome ◽  
Non Invasive

Abstract Background Non-invasive finger-cuff monitors measuring cardiac index and vascular tone (SVRI) classify emergency department (ED) patients with acute heart failure (AHF) into three otherwise-indistinguishable subgroups. Our goals were to validate these hemodynamic profiles in an external cohort and assess their association with clinical outcomes. Methods AHF patients (n=257) from five EDs were prospectively enrolled in the validation cohort (VC). Cardiac index and SVRI were measured with a ClearSight finger-cuff monitor (formerly NexFin, Edwards Lifesciences) as in a previous study (derivation cohort, DC, n=127). A control cohort (CC, n=127) of ED patients with sepsis was drawn from the same study as the DC. K-means cluster analysis previously derived two-dimensional (cardiac index and SVRI) hemodynamic profiles in the DC and CC (k=3 profiles each). The VC was subgrouped de novo into three analogous profiles by unsupervised K-means consensus clustering. PERMANOVA tested whether VC profiles 1-3 differed from profiles 1-3 in the DC and CC, by multivariate group composition of cardiac index and vascular tone. Profiles in the VC were compared by a primary outcome of 90-day mortality and a 30-day ranked composite secondary outcome (death, mechanical cardiac support, intubation, new/emergent dialysis, coronary intervention/surgery) as time-to-event (survival analysis) and binary events (odds ratio, OR). Descriptive statistics were used to compare profiles by two validated risk scores for the primary outcome, and one validated score for the secondary outcome. Results The VC had median age 60 years (interquartile range {49-67}), and was 45% (n=116) female. Multivariate profile composition by cardiac index and vascular tone differed significantly between VC profiles 1-3 and CC profiles 1-3 (p=0.001, R2=0.159). A difference was not detected between profiles in the VC vs. the DC (p=0.59, R2=0.016). VC profile 3 had worse 90-day survival than profiles 1 or 2 (HR = 4.8, 95%CI 1.4-17.1). The ranked secondary outcome was more likely in profile 1 (OR = 10.0, 1.2-81.2) and profile 3 (12.8, 1.7-97.9) compared to profile 2. Diabetes prevalence and blood urea nitrogen were lower in the high-risk profile 3 (p<0.05). No significant differences between profiles were observed for other clinical variables or the 3 clinical risk scores. Conclusions Hemodynamic profiles in ED patients with AHF, by non-invasive finger-cuff monitoring of cardiac index and vascular tone, were replicated de novo in an external cohort. Profiles showed significantly different risks of clinically-important adverse patient outcomes.

Association of Cyr61-cysteine-rich protein 61 and short-term mortality in patients with acute heart failure and coronary heart disease

2019 ◽  
Vol 13 (18) ◽  
pp. 1589-1597
Author(s):  
Chen Liu ◽  
Yalin Cao ◽  
Xin He ◽  
Chongyu Zhang ◽  
Jian Liu ◽  
...  
Keyword(s):  
Heart Failure ◽  
Coronary Heart Disease ◽  
Heart Disease ◽  
Acute Heart Failure ◽  
Logistic Models ◽  
Risk Scores ◽  
Multivariate Logistic Regression Model ◽  
Treatment Risk ◽  
All Cause Mortality ◽  
Short Term Mortality

Aim: The protein CCN1/CYR61 exerts critical functions in myocardial ischemic injury. We sought to investigate the prognostic value of CCN1 in patients with acute heart failure (AHF) and coronary heart disease (CAD). Methodology: We prospectively enrolled 113 patients with AHF and CAD. Patients were followed for all-cause mortality during a 30-day follow-up. Logistic models were used to estimate the association of CCN1 concentrations with 30-day mortality. Results: In multivariate logistic regression model, CCN1 was a significant predictor of 30-day mortality independent of current markers. Enhanced Feedback for Effective Cardiac Treatment risk score was recommended as one of the selected multivariable risk scores to predict outcome in AHF. CCN1 improved risk stratification for all-cause mortality when added to the Enhanced Feedback for Effective Cardiac Treatment risk scores at 30 days. Conclusion: We found CCN1 is independently associated with 30-day mortality in patients with AHF and CAD.

Predicting Outcome in Patients with Acute Coronary Syndrome: Evaluation of B-Type Natriuretic Peptide and the Global Registry of Acute Coronary Events (GRACE) Risk Score

2007 ◽  
Vol 52 (3) ◽  
pp. 8-13 ◽  
Author(s):  
H. Sinclair ◽  
M Paterson ◽  
S. Walker ◽  
G Beckett ◽  
K.A.A. Fox
Keyword(s):  
Heart Failure ◽  
Natriuretic Peptide ◽  
Risk Scores ◽  
Prognostic Information ◽  
Coronary Syndrome ◽  
Clinical Risk ◽  
St Elevation ◽  
Coronary Events ◽  
Global Registry ◽  
Grace Score

Background Accurate risk stratification soon after admission for patients with acute coronary syndromes (ACS) is vital in guiding management. Clinical risk scores and B-type natriuretic peptide (BNP) can predict mortality and re-infarction in ACS, but it is unknown whether BNP provides prognostic information over and above that of the clinical risk scores. Methods 142 unselected patients with ACS were prospectively studied. BNP was measured and patients were stratified according to BNP and Global Registry of Acute Coronary Events (GRACE) score. In-hospital and 30-day events were characterised. Results 20.4% of ACS subjects had ST-elevation myocardial infarction (MI), 14.1%, non-ST elevation MI and 65.5% unstable angina. Elevated BNP predicted inhospital and 30-day heart failure (p<0.01), and the risk of in-hospital recurrent ACS (p<0.05). Increasing GRACE score predicted in-hospital recurrent ACS (p<0.05), heart failure (p<0.001), arrhythmias (p<0.05) and angioplasty (p<0.05). GRACE score also predicted 30-day heart failure (p<0.05). In contrast, the predictive accuracy of troponin elevation was less robust. Conclusion BNP and the GRACE score predict complementary outcomes from ACS, but both predicted heart failure. BNP is a powerful indicator of heart failure in patients with ACS and provides prognostic information above and beyond conventional biomarkers and risk scores.

Abstract 15341: Multi-center Validation of a Novel Echocardiography Artificial Intelligence System to Predict Post-operative Right Ventricular Failure in LVAD Patients

Circulation ◽  
2020 ◽  
Vol 142 (Suppl_3) ◽  
Author(s):  
Rohan Shad Arora ◽  
Nicolas Quach ◽  
Robyn Fong ◽  
Sandra Kong ◽  
Patpilai Kasinpila ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Right Ventricular ◽  
Characteristic Curve ◽  
Left Ventricular ◽  
Right Ventricular Failure ◽  
Ventricular Assist Devices ◽  
Risk Scores ◽  
Left Ventricular Assist Devices ◽  
Ventricular Failure ◽  
Clinical Risk

Introduction: Post-operative right ventricular failure (RV failure) is the single largest contributor to short-term mortality in patients with left ventricular assist devices (LVAD); yet predicting which patient is at risk of developing this complication in the pre-operative setting has remained beyond the abilities of experts in the field. We hypothesized that deep artificial intelligence (AI) driven characterization of subtle pre-operative myocardial motion features, could predict post-operative RV failure in LVAD patients. Methods: We developed a novel echocardiography AI system using an improved dense trajectory algorithm that tracks motion vectors in an unsupervised fashion, and a 3-dimensional convolutional neural network that tracks spatiotemporal features from videos. We used pre-operative ECHO videos from a 536 patient multicenter echocardiographic and clinical dataset, and via a standard 10-fold cross validation, trained and validated the AI system to predict severe or higher grades of post-operative RV failure at the time of index hospitalization. Finally, we independently benchmarked our AI system against clinicians equipped with contemporary clinical risk scores to predict post-operative RV failure (Penn and CRITT score) and manually derived echocardiographic metrics of RV function. Results: 173 (32%) patients were adjudicated to have severe post-operative RV failure. The area under the receiver operator characteristic curve (AUC ROC) for the AI system was 0.86 (95% CI 0.824-0.896). The performance of our AI system exceeded that of both the best performing clinical risk score and manually derived echocardiographic metric by a significant margin (ΔAUC +0.30, 95% CI 0.27 - 0.33 and ΔAUC +0.24, 95% CI 0.21 - 0.27; both p &lt 0.0001). Conclusions: A novel ECHO AI system trained and validated on a multicenter dataset outperformed clinical experts equipped with both contemporary risk scores and manually calculated echocardiographic metrics.

scholarly journals Machine learning-based mortality prediction of patients undergoing cardiac resynchronization therapy: the SEMMELWEIS-CRT score

2020 ◽  
Vol 41 (18) ◽  
pp. 1747-1756 ◽  
Author(s):  
Márton Tokodi ◽  
Walter Richard Schwertner ◽  
Attila Kovács ◽  
Zoltán Tősér ◽  
Levente Staub ◽  
...  
Keyword(s):  
Machine Learning ◽  
Cardiac Resynchronization Therapy ◽  
Mortality Prediction ◽  
Cardiac Resynchronization ◽  
Risk Scores ◽  
Resynchronization Therapy ◽  
Discriminative Ability ◽  
All Cause Mortality ◽  
Estimated Risk ◽  
Stratification System

Abstract Aims Our aim was to develop a machine learning (ML)-based risk stratification system to predict 1-, 2-, 3-, 4-, and 5-year all-cause mortality from pre-implant parameters of patients undergoing cardiac resynchronization therapy (CRT). Methods and results Multiple ML models were trained on a retrospective database of 1510 patients undergoing CRT implantation to predict 1- to 5-year all-cause mortality. Thirty-three pre-implant clinical features were selected to train the models. The best performing model [SEMMELWEIS-CRT score (perSonalizEd assessMent of estiMatEd risk of mortaLity With machinE learnIng in patientS undergoing CRT implantation)], along with pre-existing scores (Seattle Heart Failure Model, VALID-CRT, EAARN, ScREEN, and CRT-score), was tested on an independent cohort of 158 patients. There were 805 (53%) deaths in the training cohort and 80 (51%) deaths in the test cohort during the 5-year follow-up period. Among the trained classifiers, random forest demonstrated the best performance. For the prediction of 1-, 2-, 3-, 4-, and 5-year mortality, the areas under the receiver operating characteristic curves of the SEMMELWEIS-CRT score were 0.768 (95% CI: 0.674–0.861; P &lt; 0.001), 0.793 (95% CI: 0.718–0.867; P &lt; 0.001), 0.785 (95% CI: 0.711–0.859; P &lt; 0.001), 0.776 (95% CI: 0.703–0.849; P &lt; 0.001), and 0.803 (95% CI: 0.733–0.872; P &lt; 0.001), respectively. The discriminative ability of our model was superior to other evaluated scores. Conclusion The SEMMELWEIS-CRT score (available at semmelweiscrtscore.com) exhibited good discriminative capabilities for the prediction of all-cause death in CRT patients and outperformed the already existing risk scores. By capturing the non-linear association of predictors, the utilization of ML approaches may facilitate optimal candidate selection and prognostication of patients undergoing CRT implantation.

Clinical phenotypes of atrial fibrillation and risk of mortality: a cluster analysis

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
D Pastori ◽  
E Antonucci ◽  
A Milanese ◽  
F Violi ◽  
P Pignatelli ◽  
...  
Keyword(s):  
Risk Factors ◽  
Heart Failure ◽  
Cluster Analysis ◽  
Risk Scores ◽  
Peripheral Artery ◽  
Clinical Phenotypes ◽  
Risk Of Mortality ◽  
All Cause Mortality ◽  
Artery Disease ◽  
Cluster 2

Abstract Background Patients with atrial fibrillation (AF) experience a high mortality rate despite optimal antithrombotic treatment. Characteristics of AF patients at higher mortality risk have been barely described so far and no risk score has been specifically developed at this aim. Furthermore, a clinical approach based on risk scores present some limits such as to not consider some important risk factors for mortality, and many available scores have poor predictive value. Cluster analysis may play a role in overcoming limitations of risk scores, especially in the case of overlapping risk factors. Purpose To identify of clinical phenotypes by using an unbiased statistical approach, such as the cluster analysis. Methods Cluster analysis was used to identify clinical phenotypes of AF patients associated with all-cause mortality in 5,171 AF patients from the START registry. Clinical variables used for the analysis were age, sex, diabetes, previous cerebrovascular events, previous cardiovascular events, heart failure, peripheral artery disease, use of non-vitamin K oral anticoagulants, cancer, pulmonary disease, smoking habit, previous major bleeding. The risk of all-cause mortality in each cluster was analyzed. Results We identified 4 clusters (Figure 1). Cluster 1 was composed by youngest patients, with obesity and paroxysmal AF; Cluster 2 by patients with low cardiovascular risk factors and high proportion of cancer; Cluster 3 by men with diabetes and coronary and peripheral artery disease, a high proportion of thrombocytopenia, and a high use of aspirin, proton pump inhibitors, and statins; Cluster 4 included the oldest patients, mainly women, with previous cerebrovascular disease, persistent/ permanent AF, heart failure, kidney disease and anemia. In this cluster there was the highest use of digoxin and NOACs. During 9856,84 patient/years of observation, 386 deaths (3.92%/year) occurred. Mortality rates significantly increased across clusters: 0.42%/year (cluster 1, reference group), 2.12%/year (cluster 2, adjusted hazard ratio [aHR] 3.306, 95% confidence interval [CI] 1.204–9.077, p=0.020), 4.41%/year (cluster 3, aHR 6.702, 95% CI 2.433–18.461, p&lt;0.001) and 8.71%/year (cluster 4, aHR 8.927, 95% CI 3.238–24.605, p&lt;0.001). Conclusions We identified different clinical phenotypes of AF patients by cluster analysis which were specifically associated with mortality. This approach may help identify patients at higher risk of mortality. Figure 1 Funding Acknowledgement Type of funding source: None

Sign in / Sign up

Export Citation Format

Share Document