EFFICIENT DEEP LEARNING FOR CHAGAS CARDIOMYOPATHYDETECTION: DATA-DRIVEN ECG FEATURE REDUCTION

Chagas cardiomyopathy, a severe complication of chronic Trypanosoma cruzi infection,requires early and efficient diagnosis to improve outcomes. While electrocardiograms (ECGs) areaccessible diagnostic tools, their high dimensionality limits the scalability of deep learning methods. Thisstudy proposes an interpretable and computationally efficient deep learning pipeline combining PrincipalComponent Analysis (PCA) for feature reduction and a Convolutional Neural Network (CNN) forclassification of ECG signals.Using data from the publicly available SaMi-Trop cohort, the ECG signals were preprocessed witha Butterworth high-pass filter to remove baseline wander, followed by PCA to reduce feature space whilepreserving critical variance. A custom CNN was then trained on the reduced feature set to classify signalsas Chagas or non-Chagas.The PCA+CNN model achieved 95.52% sensitivity, 91.11% specificity, 93.31% accuracy, and anAUC of 96.78%, outperforming the baseline CNN trained on unreduced data. The approach reducedoverfitting, improved generalization, and accelerated convergence, demonstrating strong potential for usein real-time and low-resource healthcare settings.