Our current research focuses on developing new machine learning (ML) and deep learning (DL) methods with statistical analysis from a data-driven approach using multiomics data and wearable sensor data to study multi-tissue responses to exercise and under disease model, to provide insights for therapeutic benefits. We use ML and DL methods to model the complexity and heterogeneity of multiomic data and wearable data. For multiomics data, in our ongoing work, we use network analysis and graphical neural network (GNN) to study how tissues communicate with each other in response to exercise. For the wearable data, we use signal processing and signal detection techniques and recurrent neural network (RNN) to track and evaluate the exercise responses in different individuals under various disease statuses. Finally, we will take advantage of ML and DL to link biomarkers at the molecular level with digital markers from wearable sensors to optimize training programs and to improve exercise efficiency.