Machine Learning Outcome Prediction in Dilated Cardiomyopathy Using Regional Left Ventricular Multiparametric Strain

Area of Science:

• Cardiology
• Medical Imaging
• Machine Learning

Background:

• Idiopathic dilated cardiomyopathy (IDCM) heart failure (HF) patients often present similarly, making it difficult to predict response to medical therapy.
• Distinguishing between responders and non-responders is crucial for optimizing treatment strategies and avoiding ineffective interventions.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) tool utilizing regional left ventricular (LV) contractile patterns to predict medical therapy response in IDCM HF patients.
• To identify non-responders for early, targeted interventions and prevent unnecessary surgeries in responders.

Main Methods:

• 178 subjects (140 controls, 38 IDCM patients) underwent cardiac MRI for multiparametric strain analysis.
• Longitudinal, circumferential, and radial strain were calculated across 18 LV sub-regions.
• Support vector machines (SVM), logistic regression (LR), random forest (RF), and deep neural networks (DNN) were trained to predict therapy response.

Main Results:

• The DNN model achieved the highest predictive accuracy, with an area under the receiver operating characteristic curve (AUC) of 0.94.
• Key predictive features included longitudinal strain in specific basal and mid-LV sub-regions.
• Regional contractile injury patterns effectively predicted response to medical therapy.

Conclusions:

• Regional LV contractile injury patterns derived from MRI strain analysis can accurately predict medical therapy response in IDCM HF.
• This ML-based approach shows significant potential for personalized patient care in heart failure management.