Machine learning-based risk model using 123I-metaiodobenzylguanidine to differentially predict modes of cardiac death in heart failure

Abstract Aims Heart failure (HF) is associated with an increased risk of sudden cardiac death (SCD). This study sought to demonstrate the incidence of SCD within a multicentre Japanese registry of HF patients hospitalized for acute decompensation, and externally validate the Seattle Proportional Risk Model (SPRM). Methods and results We consecutively registered 2240 acute HF patients from academic institutions in Tokyo, Japan. The discrimination and calibration of the SPRM were assessed by the c-statistic, Hosmer–Lemeshow statistic, and visual plotting among non-survivors. Patient-level SPRM predictions and implantable cardioverter-defibrillator (ICD) benefit [ICD estimated hazard ratio (HR), derived from the Cox proportional hazards model in the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT)] was calculated. During the 2-year follow-up, 356 deaths (15.9%) occurred, which included 76 adjudicated SCDs (3.4%) and 280 non-SCDs (12.5%). The SPRM showed acceptable discrimination [c-index = 0.63; 95% confidence interval (CI) 0.56–0.70], similar to that of original SPRM-derivation cohort. The calibration plot showed reasonable conformance. Among HF patients with reduced ejection fraction (EF; < 40%), SPRM showed improved discrimination compared with the ICD eligibility criteria (e.g. New York Heart Association functional Class II–III with EF ≤ 35%): c-index = 0.53 (95% CI 0.42–0.63) vs. 0.65 (95% CI 0.55–0.75) for SPRM. Finally, in the subgroup of 246 patients with both EF ≤ 35% and SPRM-predicted risk of ≥ 42.0% (SCD-HeFT defined ICD benefit threshold), mean ICD estimated HR was 0.70 (30% reduction of all-cause mortality by ICD). Conclusion The cumulative incidence of SCD was 3.4% in Japanese HF registry. The SPRM performed reasonably well in Japanese patients and may aid in improving SCD prediction.

Download Full-text

Abstract 12872: Machine Learning on Dual-isotope Myocardial Semiconductor SPECT Can Predict Cardiac Prognosis in Patients With Congestive Heart Failure

Circulation ◽

10.1161/circ.142.suppl_3.12872 ◽

2020 ◽

Vol 142 (Suppl_3) ◽

Author(s):

Masato Shimizu ◽

shummo cho ◽

Yoshiki Misu ◽

Mari Ohmori ◽

Ryo Tateishi ◽

...

Keyword(s):

Machine Learning ◽

Heart Failure ◽

Congestive Heart Failure ◽

Random Forest ◽

Roc Curve ◽

Cardiac Death ◽

Curve Analysis ◽

Roc Curve Analysis ◽

Dual Isotope ◽

Machine Learning Model

Introduction: Dual-isotope (201TlCl and 123I-β-methyl-P-iodophenyl-pentadecanoic acid (BMIPP) ) single photon emission computed tomography (SPECT) is utilized to estimate not only in patients with ischemic heart disease but with congestive heart failure (CHF). We tried to construct predictive model for cardiac prognosis on the SPECT for cardiac death by machine learning. Hypothesis: Machine learning is a powerful tool to predict cardiac prognosis in patients with CHF Methods: Consecutive 310 patients who admitted with CHF (77.1±3.1 years, 164 males) were enrolled. After initial treatment, they underwent electrocardiography gated SPECT and observed in median 507 days [IQR: 165, 1032]. Multivariate Cox regression analysis for cardiac death was performed, and predictive model was constructed by ROC curve analysis and machine learning (Random Forest and Deep Learning). The accuracies (= [True positive + True negative] / Total) of the prediction models were compared with ROC curve model. Results: Thirty-six patients fell into cardiac death. Cox analysis showed Age, left ventricular ejection fraction (LVEF), summed rest score (SRS) of BMIPP, and mismatch score were significant predictors (Hazard ratio: 1.068, 0.970, 1.032, 1.092, P value: <0.001, 0.014, 0.002, <0.001, respectively). ROC curve analysis of them revealed the accuracy of the cut-off value was 0.479-0.773. Conversely, machine learning model demonstrated higher accuracy for cardiac death (Random Forest: 0.895, Deep Learning: 0.935). The top 4 feature importance of the random forest were LVEF (0.299), SRS BMIPP (0.263), Age (0.262), and mismatch score (0.160). Conclusions: Machine learning model on SPECT had powerful predictive value for predicting cardiac death in patients with CHF.

Download Full-text

Machine learning for prediction of sudden cardiac death in heart failure patients with low left ventricular ejection fraction: study protocol for a retroprospective multicentre registry in China

BMJ Open ◽

10.1136/bmjopen-2018-023724 ◽

2019 ◽

Vol 9 (5) ◽

pp. e023724 ◽

Cited By ~ 3

Author(s):

Fanqi Meng ◽

Zhihua Zhang ◽

Xiaofeng Hou ◽

Zhiyong Qian ◽

Yao Wang ◽

...

Keyword(s):

Machine Learning ◽

Heart Failure ◽

Sudden Cardiac Death ◽

Left Ventricular Ejection Fraction ◽

Ejection Fraction ◽

Study Protocol ◽

Cardiac Death ◽

Left Ventricular ◽

Left Ventricular Ejection ◽

Ventricular Ejection Fraction

IntroductionLeft ventricular ejection fraction (LVEF) ≤35%, as current significant implantable cardioverter-defibrillator (ICD) indication for primary prevention of sudden cardiac death (SCD) in heart failure (HF) patients, has been widely recognised to be inefficient. Improvement of patient selection for low LVEF (≤35%) is needed to optimise deployment of ICD. Most of the existing prediction models are not appropriate to identify ICD candidates at high risk of SCD in HF patients with low LVEF. Compared with traditional statistical analysis, machine learning (ML) can employ computer algorithms to identify patterns in large datasets, analyse rules automatically and build both linear and non-linear models in order to make data-driven predictions. This study is aimed to develop and validate new models using ML to improve the prediction of SCD in HF patients with low LVEF.Methods and analysisWe will conduct a retroprospective, multicentre, observational registry of Chinese HF patients with low LVEF. The HF patients with LVEF ≤35% after optimised medication at least 3 months will be enrolled in this study. The primary endpoints are all-cause death and SCD. The secondary endpoints are malignant arrhythmia, sudden cardiac arrest, cardiopulmonary resuscitation and rehospitalisation due to HF. The baseline demographic, clinical, biological, electrophysiological, social and psychological variables will be collected. Both ML and traditional multivariable Cox proportional hazards regression models will be developed and compared in the prediction of SCD. Moreover, the ML model will be validated in a prospective study.Ethics and disseminationThe study protocol has been approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (2017-SR-06). All results of this study will be published in international peer-reviewed journals and presented at relevant conferences.Trial registration numberChiCTR-POC-17011842; Pre-results.

Download Full-text

A novel method to predict the proportional risk of sudden cardiac death in heart failure: Derivation of the Seattle Proportional Risk Model

Heart Rhythm ◽

10.1016/j.hrthm.2015.06.039 ◽

2015 ◽

Vol 12 (10) ◽

pp. 2069-2077 ◽

Cited By ~ 45

Author(s):

Ramin Shadman ◽

Jeanne E. Poole ◽

Todd F. Dardas ◽

Dariush Mozaffarian ◽

John G.F. Cleland ◽

...

Keyword(s):

Heart Failure ◽

Sudden Cardiac Death ◽

Cardiac Death ◽

Risk Model ◽

Novel Method

Download Full-text

Personalized prediction of mode of cardiac death in heart failure using supervised machine learning in the context of cardiac innervation imaging

Journal of Nuclear Cardiology ◽

10.1007/s12350-020-02215-z ◽

2020 ◽

Author(s):

Rudolf A. Werner ◽

Thorsten Derlin ◽

Frank M. Bengel

Keyword(s):

Machine Learning ◽

Heart Failure ◽

Cardiac Death ◽

Supervised Machine Learning ◽

Cardiac Innervation

Download Full-text

Sudden cardiac death and pump failure death prediction in chronic heart failure by combining ECG and clinical markers in an integrated risk model

PLoS ONE ◽

10.1371/journal.pone.0186152 ◽

2017 ◽

Vol 12 (10) ◽

pp. e0186152 ◽

Cited By ~ 16

Author(s):

Julia Ramírez ◽

Michele Orini ◽

Ana Mincholé ◽

Violeta Monasterio ◽

Iwona Cygankiewicz ◽

...

Keyword(s):

Heart Failure ◽

Sudden Cardiac Death ◽

Chronic Heart Failure ◽

Cardiac Death ◽

Risk Model ◽

Clinical Markers ◽

Pump Failure ◽

Integrated Risk

Download Full-text

Prediction for cardiac prognosis in patients with congestive heart failure by machine learning on dual-isotope myocardial semiconductor SPECT

European Heart Journal ◽

10.1093/ehjci/ehaa946.0275 ◽

2020 ◽

Vol 41 (Supplement_2) ◽

Author(s):

M Shimizu ◽

S Cho ◽

K Hara ◽

M Ohmori ◽

R Tateishi ◽

...

Keyword(s):

Machine Learning ◽

Heart Failure ◽

Random Forest ◽

Roc Curve ◽

Cardiac Death ◽

Learning Model ◽

Curve Analysis ◽

Roc Curve Analysis ◽

Dual Isotope ◽

Machine Learning Model

Abstract Background Dual-isotope (low doze 201TlCl and 123I-β-methyl-P-iodophenyl-pentadecanoic acid (BMIPP)) single photon emission computed tomography (SPECT) is utilized to estimate myocardial damage in patients with congestive heart failure (CHF). However, predictive model construction on the SPECT for cardiac death by machine learning was not studied. Purpose To elucidate predictive value of machine learning model on dual-isotope SPECT for CHF. Methods We enrolled consecutive 310 patients who admitted with CHF (77.1±3.1 years, 164 males). After initial treatment, they underwent electrocardiography gated SPECT and observed in median 507 days [IQR: 165, 1032]. Multivariate Cox regression analysis for cardiac death was performed, and predictive model was constructed by ROC curve analysis and machine learning (Random Forest and Deep Learning). The accuracies (= [True positive + True negative] / Total) of the prediction models were compared with ROC curve model. Results Thirty-six patients fell into cardiac death. Cox analysis showed Age, left ventricular ejection fraction (LVEF), summed rest score (SRS) of BMIPP, and mismatch score were significant predictors (Hazard ratio: 1.068, 0.970, 1.032, 1.092, P value: <0.001, 0.014, 0.002, <0.001, respectively). ROC curve analysis of them revealed the accuracy of the cut-off value was 0.479–0.773. Conversely, machine learning model demonstrated higher accuracy for cardiac death (Random Forest: 0.895, Deep Learning: 0.935). The top 4 feature importance of the random forest were LVEF (0.299), SRS BMIPP (0.263), Age (0.262), and mismatch score (0.160). Conclusion Machine learning model on SPECT was superior to conventional statistic model for predicting cardiac death in patients with CHF. Funding Acknowledgement Type of funding source: None

Download Full-text

Creation of mortality risk calculator using a I-123 mIBG-based machine learning model: differential prediction of arrhythmic death and heart-failure death

European Heart Journal - Cardiovascular Imaging ◽

10.1093/ehjci/jeab111.055 ◽

2021 ◽

Vol 22 (Supplement_3) ◽

Author(s):

K Nakajima ◽

T Nakata ◽

T Doi ◽

H Tada ◽

S Saito ◽

...

Keyword(s):

Machine Learning ◽

Heart Failure ◽

Mortality Risk ◽

Cardiac Death ◽

Regression Method ◽

Risk Calculator ◽

Differential Prediction ◽

Arrhythmic Death ◽

Logistic Regression Method ◽

Low Probability

Abstract Funding Acknowledgements Type of funding sources: None. Background Although I-123 meta-iodobenzylguanidine (mIBG) has been applied to patients with chronic heart failure (CHF), a diagnostic tool for differential prediction of fatal arrhythmic events (ArE) and heart-failure death (HFD) has been pursued. Purpose The aim of this study was to create a calculator of mortality risk for differentiating mode of cardiac death using a machine learning (ML) method, and to test the accuracy in a new cohort of patients with CHF. Methods A total of 529 patients with CHF was used as the training database for ML. The ArE group consisted of patients with arrhythmic death, sudden cardiac death and appropriate therapy by implantable cardioverter defibrillator. A heart-to-mediastinum ratio (H/M) standardized to the medium-energy collimator condition was calculated with a planar anterior mIBG scintigram. The best classifier models for predicting HFD and ArE were determined by four-fold cross validation. Input variables included age, sex, New York Heart Association (NYHA) functional class, left ventricular ejection fraction, ischemic etiology, mIBG H/M and washout rate, and b-type natriuretic peptide (BNP) or NT Pro BNP, estimated glomerular filtration rate, hemoglobin, and complications such as diabetes and hypertension. After creating the ML-based model, the constructed classifier functions for ArE, HFD, and survival were exported for subsequent use. A new cohort of patients (n = 312, age 67 ± 13 years, 2015 or later) was used to test the ML-based model. Results The training database included 141 events (27%) with ArE (7%) and HFD (20%). Receiver-operating characteristic analysis by four-fold validation showed area under the curve value of 0.90 for HFD and 0.73 for ArE. Among various ML methods, the logistic regression method demonstrated the most stable calculation of the probability of ArE followed by random forest and gradient boosted tree methods. Therefore, the logistic-regression method was used for calculating both HFD and ArE probabilities. In the test cohort, patients with a high HFD probability >8% resulted in 6.3-fold higher HFD than those with low probability (≤ 8%). Patients with high ArE probability >8% showed 2.5-fold higher ArE than those with low probability (≤ 8%). Conclusion The ML-based mortality risk calculator could be used for stratifying patients at high and low risks, which might be useful for estimating appropriate treatment strategy.

Download Full-text