Chagas disease, a neglected tropical disease affecting over 6 million people globally, often presents with subtle ECG abnormalities, complicating diagnosis. Chagas disease is a leading cause of cardiovascular morbidity, yet accurate detection remains challenging. This study proposes two computational approaches, machine learning and deep learning, for classifying Chagas and non-Chagas disease using ECG signals. A multi-scale one-dimensional convolutional neural network (CNN) was developed, incorporating three convolutional blocks, global average pooling, and focal loss to address class imbalance, achieving robust binary classification across 12-lead ECGs normalized to 5000 samples. Additionally, a random forest (RF) classifier was trained on extracted features, including morphological and time-series attributes from P, QRS, and T waves, after denoising with a bandpass filter. Both methods were evaluated on a Physionet dataset, with the CNN demonstrating high accuracy and confidence calibration. These scalable techniques provide promising tools for automated Chagas disease detection, supporting precision cardiology and enhancing clinical outcomes. Moreover, it highlights the effectiveness of the CNN model in detecting Chagas disease using the focal loss function. Our team, Leicester Fox, had a challenge score of 0.258 for an official phase using the CNN classification model.