Machine Learning Prediction of Response to Cardiac Resynchronization Therapy

Feeny, Albert; Rickard, John; Patel, Divyang; Toro, Saleem; Trulock, Kevin; Park, Carolyn; LaBarbera, Michael A.; Varma, Niraj; Niebauer, Mark; Sinha, Sunil K.; Gorodeski, Eiran Z.; Grimm, Richard A.; Ji, Xinge; Barnard, John; Madabhushi, Anant; Spragg, David D.; Chung, Mina K.

doi:10.1161/circep.119.007316

Cited by 87 publications

(86 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…ML can improve the care of HF patients in various ways, e.g., by augmenting the prediction of readmission after HF hospitalization or by predicting the risk of mortality (16,17,19). In HF patients undergoing CRT implantation, our research group has previously confirmed the superiority of ML over pre-existing risk scores (24), and similar results have been reported by others as well (25,26). Underpinning these findings, we were able to predict the 1-and 3-year mortality of CRT patients with good discrimination and excellent calibration, even in subsets of patients divided by sex.…”

Section: Risk Stratification Of Hf Patients Using MLsupporting

confidence: 81%

Sex-Specific Patterns of Mortality Predictors Among Patients Undergoing Cardiac Resynchronization Therapy: A Machine Learning Approach

Tokodi

Behon

Merkel

et al. 2021

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

Background: The relative importance of variables explaining sex-related differences in outcomes is scarcely explored in patients undergoing cardiac resynchronization therapy (CRT). We sought to implement and evaluate machine learning (ML) algorithms for the prediction of 1- and 3-year all-cause mortality in CRT patients. We also aimed to assess the sex-specific differences in predictors of mortality utilizing ML.Methods: Using a retrospective registry of 2,191 CRT patients, ML models were implemented in 6 partially overlapping patient subsets (all patients, females, or males with 1- or 3-year follow-up). Each cohort was randomly split into training (80%) and test sets (20%). After hyperparameter tuning in the training sets, the best performing algorithm was evaluated in the test sets. Model discrimination was quantified using the area under the receiver-operating characteristic curves (AUC). The most important predictors were identified using the permutation feature importances method.Results: Conditional inference random forest exhibited the best performance with AUCs of 0.728 (0.645–0.802) and 0.732 (0.681–0.784) for the prediction of 1- and 3-year mortality, respectively. Etiology of heart failure, NYHA class, left ventricular ejection fraction, and QRS morphology had higher predictive power, whereas hemoglobin was less important in females compared to males. The importance of atrial fibrillation and age increased, while the importance of serum creatinine decreased from 1- to 3-year follow-up in both sexes.Conclusions: Using ML techniques in combination with easily obtainable clinical features, our models effectively predicted 1- and 3-year all-cause mortality in CRT patients. Sex-specific patterns of predictors were identified, showing a dynamic variation over time.

show abstract

Section: Risk Stratification Of Hf Patients Using MLsupporting

confidence: 81%

Sex-Specific Patterns of Mortality Predictors Among Patients Undergoing Cardiac Resynchronization Therapy: A Machine Learning Approach

Tokodi

Behon

Merkel

et al. 2021

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

show abstract

“…The predictive value of current class 1 indications for CRT response is modest and little improved by machine learning, with area under the curve (AUC) of 0.65 and 0.70, respectively. 19 This is similar for qLV >95 ms or RVs/LVs, which correlates with qLV. 20 However, these criteria omit measures of successful delivery.…”

Section: Discussionmentioning

confidence: 67%

Left ventricular paced activation in cardiac resynchronization therapy patients with left bundle branch block and relationship to its electrical substrate

Wisnoskey

Varma

2020

Heart Rhythm O2

Self Cite

View full text Add to dashboard Cite

BACKGROUND Cardiac resynchronization therapy (CRT) uses left ventricular (LV) pacing to restore rapid synchronized LV activation when it is delayed in patients with myocardial disease. OBJECTIVE Although intrinsic LV activation delays are understood, little is known about reactions to LV stimulation and whether they are affected by QRS duration (QRSd), morphology, LV substrate, or choice of electrode pair. The purpose of this study was to test these interactions. METHODS In 120 heart failure patients with left bundle branch block (LBBB) and QRS .120 ms receiving CRT with quadripolar LV leads, device-based measurements of intrinsic activation delay (qLV) and paced inter-(and intra-) LV conduction times were evaluated at the proximal and distal LV bipoles. RESULTS During intrinsic conduction, qLV varied little between the proximal and distal pairs in patients with LBBB (n 5 120; age 68 6 11 years; 63% male; ejection fraction 25% 6 7%; 33% ischemic cardiomyopathy; QRSd 162 6 19 ms). A minority (30%) had conduction barriers (ie, gradients) (DqLV 29 6 8 ms vs 9 6 5 ms in patients without gradients; P ,.01), which occurred equally in ischemic and nonischemic patients. A majority were functional (and not scar-mediated), as they resolved with pacing in most patients (75%). Importantly, LV-paced conduction times were unrelated to baseline QRS morphology (LBBB 166 6 30 ms vs RBBB control 172 6 30 ms; P 5 NS), longer than intrinsic conduction (166 6 30 ms vs 129 6 28 ms; P ,.01), and varied significantly by electrode pair (ie, small distances) and etiology. Correlation between intrinsic activation delay (qLV) and LV-paced conduction time was poor (R 2 5 0.278; P ,.05). CONCLUSION LV-paced effect, which is core to CRT, is unpredictable based on conventionally used measures and should be considered during CRT optimization.

show abstract

“…Steroid therapy, when started simultaneously with CRT, might have altered the clinical course in this patient. Recently, machine learning using the patients' baseline characteristics of the patients has been applied to the prediction of response to CRT, but the results are not necessarily very remarkable to date [43,44]. A composite evaluation of machine learning scores and simulation results would potentiate the power of these two approaches.…”

Section: Discussionmentioning

confidence: 99%

Patient-specific heart simulation can identify non-responders to cardiac resynchronization therapy

et al. 2020

View full text Add to dashboard Cite

To identify non-responders to cardiac resynchronization therapy (CRT), various biomarkers have been proposed, but these attempts have not been successful to date. We tested the clinical applicability of computer simulation of CRT for the identification of non-responders. We used the multi-scale heart simulator "UT-Heart," which can reproduce the electrophysiology and mechanics of the heart based on a molecular model of the excitation-contraction mechanism. Patient-specific heart models were created for eight heart failure patients who were treated with CRT, based on the clinical data recorded before treatment. Using these heart models, bi-ventricular pacing simulations were performed at multiple pacing sites adopted in clinical practice. Improvement in pumping function measured by the relative change of maximum positive derivative of left ventricular pressure (%ΔdP/dt max) was compared with the clinical outcome. The operators of the simulation were blinded to the clinical outcome. In six patients, the relative reduction in end-systolic volume exceeded 15% in the follow-up echocardiogram at 3 months (responders) and the remaining two patients were judged as non-responders. The simulated %ΔdP/dt max at the best lead position could identify responders and non-responders successfully. With further refinement of the model, patient-specific simulation could be a useful tool for identifying non-responders to CRT.

show abstract

Machine Learning Prediction of Response to Cardiac Resynchronization Therapy

Cited by 87 publications

References 34 publications

Sex-Specific Patterns of Mortality Predictors Among Patients Undergoing Cardiac Resynchronization Therapy: A Machine Learning Approach

Sex-Specific Patterns of Mortality Predictors Among Patients Undergoing Cardiac Resynchronization Therapy: A Machine Learning Approach

Left ventricular paced activation in cardiac resynchronization therapy patients with left bundle branch block and relationship to its electrical substrate

Patient-specific heart simulation can identify non-responders to cardiac resynchronization therapy

Contact Info

Product

Resources

About