Table 1.

Steps of feature selection and the list of clinical features included in the machine learning models.

	Demographics and clinical characteristics	Comorbidities	ECG	Laboratory parameters	Medications
Included in the ML models	Age at CRT implantation Sex Body mass index NYHA functional class HF duration >18 months Etiology of heart failure LVEF LV end-diastolic diameter Type of implanted device LV lead position	Hypertension Diabetes mellitus Type of AFCOPD Smoking status Valvular heart disease	QRS morphology	Hemoglobin Serum sodium Serum creatinine	ACE-I/ARB Beta-blockers CCB Loop diuretics Thiazide diuretics MRA Digitalis Amiodarone Statin Allopurinol
Excluded due to collinearity	Height Weight	History of MI History of CABG and/or PCI		Serum urea GFR	Oral anticoagulants
Excluded due to ≥40% missing values	Systolic blood pressure Diastolic blood pressure Heart rate LV end-diastolic volume LV end-systolic volume		QRS duration PR interval	Lymphocyte Total cholesterol Serum uric acid NT-proBNP

Feature selection included two consecutive steps. First, features missing in more than 40% of patients were excluded. Then, collinear variables (Spearman correlation coefficient ≥0.3 or ≤-0.3) were also eliminated as highly correlated variables might bias the further steps of the analysis. The final set of features included 30 clinical variables: age at CRT implantation, sex, body mass index, New York Heart Association functional class, heart failure duration >18 months, etiology of heart failure (ischemic or non-ischemic), left ventricular ejection fraction and end-diastolic diameter assessed with two-dimensional echocardiography, type of the implanted device (CRT-P or CRT-D), left ventricular lead position (anterior, lateral or posterior), hypertension, diabetes mellitus, type of atrial fibrillation (paroxysmal, persistent or permanent), chronic obstructive pulmonary disease, smoking status, valvular heart disease (moderate to severe aortic valve disease, moderate to severe mitral valve disease, severe tricuspid regurgitation), QRS morphology (non-LBBB or LBBB), hemoglobin concentration, serum sodium and creatinine, medical treatment with angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, beta-blockers, calcium channel blockers, loop diuretics, thiazide diuretics, mineralocorticoid receptor antagonists, digitalis, amiodarone, statins, and allopurinol.

ACE-I, angiotensin-converting enzyme inhibitors; AF, atrial fibrillation; ARB, angiotensin II receptor blockers; CABG, coronary artery bypass grafting; CCB, calcium channel blocker; COPD, chronic obstructive pulmonary disease; CRT, cardiac resynchronization therapy; ECG, electrocardiogram; EF, ejection fraction; GFR, glomerular filtration rate; HF, heart failure; LV, left ventricular; MI, myocardial infarction; ML, machine learning; NYHA, New York Heart Association; NT-proBNP, N-terminal pro-brain natriuretic peptide; MRA, mineralocorticoid receptor antagonists; PCI, percutaneous coronary intervention.