Human Activity Recognition (HAR) has gained significant attention in recent years due to its applications in healthcare, smart homes, fitness tracking, and human-computer interaction. This study focuses on the design and evaluation of predictive models for HAR using deep learning algorithms. By leveraging sensor data from wearable devices, the proposed models aim to classify human activities accurately and efficiently. Deep learning architectures such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and hybrid models are explored to capture spatial and temporal features inherent in HAR datasets. The study utilizes publicly available datasets, such as the UCI HAR and WISDM datasets, for training and testing the models. Comprehensive experiments are conducted to evaluate the performance of these models in terms of accuracy, precision, recall, and computational efficiency.