Aims: This work aims to improve the accurate detection of Chagas Disease from ECG signals, ensuring early diagnosis and treatment, by extracting relevant features from ECG recordings and applying ensemble Machine Learning techniques. Methods: Our approach combines several feature extraction algorithms, including time-domain, frequency-domain, wavelet, and heart rate variability features. These features capture different aspects of ECG signals, allowing for a more robust representation of underlying cardiac activity. We extract features from individual ECG leads, patient demographic data (sex and age), and inter-lead correlation analysis, all used to train a Voting Ensemble model combining Random Forest and XGBoost optimized for imbalanced datasets. Results: The methodology developed by our team, Logia_DC_UFC, achieved an Unofficial Phase Leaderboard score of 0.812 (6th place on ranking). On development environment, after a 5-fold cross-validation with only 80% of the provided Base Code 15% data, we achieved the following average metrics: 94.85% Accuracy, 85.04% AUC-ROC, 17.27% AUPRC, 23.36% F-measure, and 0.414 Challenge Score. Conclusion: This work demonstrates the potential of combining multiple ECG feature extraction with ensemble learning techniques to detect Chagas Disease from ECG signals. Continuing our work will focus on extracting more clinically relevant features (such as those derived from the QRS complex), implementing more complex models, and developing better strategies to address class imbalance. Our promising results highlight the effectiveness of our multi-domain feature extraction approach and voting ensemble methodology, enabling more accurate and timely detection of Chagas Disease from standard ECG recordings.