Biomedical and healthcare systems are inherently complex, nonlinear, time-varying, and characterized by tightly coupled feedback loops, high inter- and intra-individual variability, and the frequent need for closed-loop experimentation. Addressing these challenges requires a broad methodological toolkit: classical data-driven approaches—such as frequency response estimation, subspace methods, and closed-loop identification—provide interpretable uncertainty bounds essential for clinical reliability, while machine learning and hybrid methods complement these by capturing complex nonlinear dynamics and enabling personalized modeling from heterogeneous clinical data. This Open Invited Track brings together experts in system identification and machine learning with experience or interest in biomedical and human health applications, spanning theoretical developments and experimental human studies, with the goal of highlighting recent advances, identifying open challenges, and fostering new directions in developing reliable and clinically actionable models of biomedical systems.